JP6745496B2 - Diabetic retinopathy stage determination support system and method for supporting stage determination of diabetic retinopathy - Google Patents
Diabetic retinopathy stage determination support system and method for supporting stage determination of diabetic retinopathy Download PDFInfo
- Publication number
- JP6745496B2 JP6745496B2 JP2016161579A JP2016161579A JP6745496B2 JP 6745496 B2 JP6745496 B2 JP 6745496B2 JP 2016161579 A JP2016161579 A JP 2016161579A JP 2016161579 A JP2016161579 A JP 2016161579A JP 6745496 B2 JP6745496 B2 JP 6745496B2
- Authority
- JP
- Japan
- Prior art keywords
- diabetic retinopathy
- image data
- diameter
- fundus
- range
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Description
本発明は、糖尿病網膜症の病期判定支援システムおよび糖尿病網膜症の病期の判定を支援する方法に関する。より詳細に、撮影された被験者の眼底写真から糖尿病網膜症の病期を簡易的且つ定量的に評価することができる、糖尿病網膜症の病期判定支援システムおよび糖尿病網膜症の病期の判定を支援する方法に関する。 TECHNICAL FIELD The present invention relates to a stage determination support system for diabetic retinopathy and a method for supporting stage determination of diabetic retinopathy. More specifically, it is possible to easily and quantitatively evaluate the stage of diabetic retinopathy from the photographed fundus photograph of the subject, a stage determination support system for diabetic retinopathy, and the stage determination of diabetic retinopathy. Regarding how to help.
糖尿病網膜症の重症度分類は患者管理や治療方針を決めるうえでの根拠の一つとなる。また、内科医と眼科医が診療情報や治療計画を共有して医療連携を行ううえで、共通言語として重要な意味を有しており、両者が網膜症の重症度分類に対して共通の認識を持つことが望ましい。日本では、Scott分類，改変Davis分類，新福田分類，ETDRS分類，国際重症度分類などが使用されている。
眼底検査によって糖尿病網膜症の病期を、単純糖尿病網膜症（simple diabetic retinopathy）、前増殖糖尿病網膜症（preproliferative diabetic retinopathy）、および増殖糖尿病網膜症（proliferative diabetic retinopathy）の三段階に分けることが一般的に行われている。
単純糖尿病網膜症の段階では、針の先で突いたような小さな点状出血、それよりやや大きめの斑状出血、毛細血管が膨らんでできる毛細血管瘤、脂肪やたんぱく質が沈着してできたシミ(硬性白斑)、血管がつまってできたシミ(軟性白斑)などが眼底所見として見える。この段階では、視力に全く影響なく、血糖コントロールをよくしていると自然に治癒する。
前増殖糖尿病網膜症の段階では、軟性白斑というシミが多数出てきたり、血管がつまって酸素欠乏になった部分があちこちに出てきたりする。静脈が異常に腫れ上がったり、毛細血管の形が不規則になったりすることもある。この段階は、視力に影響ないが、危険な状態に一歩踏み込んだ状態である。この段階でレーザー光凝固などの処置を行うと自然治癒しやすい。
増殖糖尿病網膜症の段階では、新生血管(正常ではないはずの新しい血管が硝子体にのびてくる)、新生血管が破れて起こる硝子体出血、増殖膜、網膜剥離が生じる。新生血管が生じてもまだ自覚症状はない。新生血管が生じただけの段階であればレーザー光凝固を行えば網膜症は治癒することが多い。しかし、硝子体出血や網膜剥離を起こしていると自然治癒は難しい。硝子体出血や網膜剥離を生じた段階になると、目の中に煙のすすがたくさん出たり、赤いカーテンがかかったりなどの自覚症状が出てくる。
Severity classification of diabetic retinopathy is one of the grounds for deciding patient management and treatment policy. In addition, it has an important meaning as a common language for physicians and ophthalmologists to share medical information and treatment plans for medical cooperation, and both have a common recognition for the severity classification of retinopathy. It is desirable to have. In Japan, Scott classification, modified Davis classification, Shin-Fukuda classification, ETDRS classification, international severity classification, etc. are used.
Fundus examination generally divides the stage of diabetic retinopathy into three stages: simple diabetic retinopathy, preproliferative diabetic retinopathy, and proliferative diabetic retinopathy. Is done in a regular manner.
At the stage of simple diabetic retinopathy, small petechial hemorrhage, which looks like a needle stick, slightly larger ecchymosis, capillary swelling caused by swelling of capillaries, and spots formed by deposition of fat and protein ( Hard fundus) and spots formed by blood vessels (soft vitiligo) are visible as fundus findings. At this stage, eyesight is not affected at all, and if blood glucose control is well done, it will be cured naturally.
At the stage of pre-proliferative diabetic retinopathy, many spots called soft vitiligo appear, and oxygen-deficient areas appear due to clogged blood vessels. The veins may be swollen abnormally or the capillaries may be irregularly shaped. This stage does not affect eyesight, but is a step into a dangerous state. If a treatment such as laser photocoagulation is performed at this stage, natural healing is likely to occur.
In the stage of proliferative diabetic retinopathy, new blood vessels (new blood vessels that should not be normal extend to the vitreous body), vitreous hemorrhage caused by breaking of new blood vessels, proliferative membrane, and retinal detachment occur. Even if new blood vessels occur, there are no subjective symptoms. Retinopathy is often cured by laser photocoagulation at the stage where only new blood vessels are generated. However, spontaneous healing is difficult when vitreous hemorrhage or retinal detachment occurs. At the stage of vitreous hemorrhage and detachment of the retina, subjective symptoms such as lots of smoke soot in the eyes and red curtains appear.
眼底検査において一般に用いられている眼底カメラは、撮影画角が４５度程度のものである。撮影画角２００度の超広角眼底カメラも存在するが高価である。撮影画角４５度の眼底カメラによる検査では、黄斑部と視神経乳頭の中央が画像中心になるように位置合わせをして眼底写真を撮影する。撮影画角４５度の眼底写真には全眼底（画角約２６０度）のうちのほんの一部が描出されるにすぎない。そのため、撮影画角の外側の範囲の眼底に病変があっても、見落とすことになる。
眼底写真には乳頭、黄斑(中心窩)、眼底動脈（網膜中心動脈、鼻側動脈、耳側動脈）、眼底静脈が描出される。通常、眼底写真に描出された血管を医師が目視で観察し、新生血管が生じているかどうかで、増殖網膜症かそうでないかの判断を行う。眼底写真には新生血管以外の血管が描出されているので、描出された血管が新生血管であるか否かの判断は、ベテラン医師でも難しく、経験の少ない医師においては見落とすこともある。
A fundus camera generally used in fundus examination has a photographing field angle of about 45 degrees. There is an ultra-wide-angle fundus camera with a shooting angle of view of 200 degrees, but it is expensive. In the examination using a fundus camera with a photographing angle of view of 45 degrees, a fundus photograph is taken by aligning the macula and the center of the optic disc so that they are in the center of the image. Only a part of the entire fundus (angle of view of about 260 degrees) is depicted in a fundus photograph with a photographing angle of view of 45 degrees. Therefore, even if there is a lesion on the fundus outside the photographing angle of view, it is overlooked.
On the fundus photograph, the papilla, macula (fovea), fundus arteries (central retinal artery, nasal artery, ear artery) and fundus vein are visualized. Usually, a doctor visually observes a blood vessel drawn on a fundus photograph to determine whether a proliferative retinopathy or not depending on whether a new blood vessel is generated. Since blood vessels other than new blood vessels are visualized in the fundus photograph, it is difficult for even an experienced doctor to judge whether or not the depicted blood vessel is a new blood vessel, and a doctor with little experience may overlook it.
眼底写真の画像データに基く病変の判断を簡便に定量的に行うために、種々の画像処理システムが提案されている。 Various image processing systems have been proposed in order to easily and quantitatively judge a lesion based on image data of a fundus photograph.
例えば、特許文献１は、被験者の眼底を撮影した眼底画像データを取得し、解析コンピュータを利用して前記眼底画像データの解析を実行することにより、眼底領域の毛細血管瘤を強調する眼底画像解析システムであって、前記解析コンピュータは、眼底カメラを利用して取得された前記被験者の前記眼底画像データの入力を受付ける眼底画像データ入力受付手段と、前記眼底画像データの前記眼底領域を構成するそれぞれの画素に対して適用され、フィルタ処理対象の注目画素を中心とする内輪領域に設定される第一フィルタ、前記第一フィルタの外周を取り囲むようにして設定される空白部、及び前記空白部の外周を取り囲む外郭領域に設定される第二フィルタを有する二重リングフィルタを形成するフィルタ形成手段と、前記眼底画像データに対し、前記二重リングフィルタの前記第一フィルタに基づいて、前記注目画素を中心とする前記内輪領域の各画素の平均値を算出する平均値算出手段と、前記眼底画像データに対し、前記二重リングフィルタの前記第二フィルタに基づいて、前記外郭領域の画素値のヒストグラムを作成するヒストグラム作成手段と、作成された前記ヒストグラムの全ヒストグラム領域の面積に対する、算出された前記平均値以上のヒストグラム領域の面積の割合を算出し、算出された割合値を前記二重リングフィルタによる出力値として生成する出力値生成手段とを具備することを特徴とする眼底画像解析システムを開示している。 For example, Patent Document 1 obtains fundus image data obtained by photographing a fundus of a subject, and analyzes the fundus image data by using an analysis computer to analyze a fundus image for emphasizing a capillary aneurysm in a fundus region. In the system, the analysis computer, a fundus image data input receiving means for receiving an input of the fundus image data of the subject obtained by using a fundus camera, and configuring the fundus region of the fundus image data, respectively. Of the first filter, which is applied to the pixel of No. 1 and is set in the inner ring region centered on the pixel of interest to be filtered, the blank portion set to surround the outer periphery of the first filter, and Filter forming means for forming a double ring filter having a second filter set in an outer peripheral region surrounding the outer circumference, and for the fundus image data, based on the first filter of the double ring filter, the pixel of interest An average value calculating means for calculating an average value of each pixel of the inner ring region centered on the, and the fundus image data, based on the second filter of the double ring filter, of the pixel value of the outer region A histogram creating means for creating a histogram, and a ratio of the area of the histogram area that is equal to or greater than the calculated average value to the area of all the histogram areas of the created histogram, and the calculated ratio value is the double ring. Disclosed is a fundus image analysis system, comprising: an output value generation unit that generates an output value by a filter.
特許文献２は、画像における（複数の）病変の有無を評定するシステムであって、ａ）画像の少なくとも１つのサブセットを評価するアルゴリズムを備え、それによって、各サブセットは、所定の可視度を有する病変候補部位であり、上記アルゴリズムは、上記病変候補部位を包囲する部位における画像の背景の変動を評価し、ｂ）上記病変候補部位の可視度を背景の変動によって補正し、上記補正された可視度を、その部位における病変の予め決められた可視度のしきい値と比較すること、又は、予め決められた可視度のしきい値を背景の変動によって補正し、上記病変候補部位の可視度を上記予め決められかつ補正された可視度のしきい値と比較することを行うアルゴリズムと、ｃ）ａ）で検出された上記病変候補部位を、ステップｂ）で取得された上記しきい値に関連して病変として、又は病変でないものとして分類するアルゴリズムと、ｄ）オプションで、すべての病変候補部位が分類されるまでステップａ）乃至ｃ）を反復するアルゴリズムとを備えたシステムを開示している。 US Pat. No. 6,037,898 is a system for assessing the presence or absence of lesion(s) in an image, comprising: a) an algorithm for assessing at least one subset of the image, whereby each subset has a predetermined visibility. The lesion candidate site, and the algorithm evaluates the variation of the background of the image in the region surrounding the lesion candidate site, and b) the visibility of the lesion candidate site is corrected by the variation of the background, and the corrected visibility The degree of visibility of the lesion candidate site by comparing the degree with a predetermined visibility threshold of the lesion at that site, or by correcting the predetermined visibility threshold with background variations. Comparing the above-mentioned predetermined and corrected threshold of visibility with c), the lesion candidate site detected in c) a) to the threshold acquired in step b). Disclosed is a system with an algorithm to classify as related or non-lesion and d) optionally repeat steps a) to c) until all candidate lesion sites have been classified. There is.
非特許文献１は、眼底写真の血管輪郭線をファジィクラスタリングを用いて抽出し、この輪郭線により血管の状態を数値化する方法を開示している。
非特許文献２は、網膜症が原因で生じた出血箇所が、眼底写真において小さな円形のパターンとして表れることを機械学習させることにより、それと同じようなパターンを持つ画像（出血箇所）を検出する方法を開示している。
上記先行技術文献にて開示されるシステムまたは方法は、眼底写真に描出された眼底血管のうちに新生血管が在るか否かなどで主に評価を行っている。眼底血管は低い明度で且つ細いスジで描出されるので、見落としやすい。
Non-Patent Document 1 discloses a method of extracting a blood vessel contour line of a fundus photograph using fuzzy clustering, and digitizing the state of the blood vessel by this contour line.
Non-Patent Document 2 is a method of detecting an image (bleeding site) having a similar pattern by machine learning that a bleeding site caused by retinopathy appears as a small circular pattern in a fundus photograph. Is disclosed.
The system or method disclosed in the above-mentioned prior art document mainly evaluates whether or not a new blood vessel exists in the fundus blood vessels depicted in the fundus photograph. The blood vessels in the fundus of the eye have low brightness and are depicted by thin stripes, so they are easy to overlook.
本発明の課題は、眼底写真により糖尿病網膜症の病期を簡易的且つ定量的に評価することができる、糖尿病網膜症の病期判定支援システムおよび病期判定支援方法を提供することである。 An object of the present invention is to provide a stage determination support system and a stage determination support method for diabetic retinopathy, which enables simple and quantitative evaluation of the stage of diabetic retinopathy by fundus photography.
上記課題を解決すべく鋭意検討した結果、下記の形態を包含する本発明を完成するに至った。 As a result of intensive studies to solve the above problems, the present invention including the following modes has been completed.
〔１〕 糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを含むデータ群に基いて、機械学習アルゴリズムを用いて、糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを写像する関数を生成させ、
被験者の眼底写真の画像データから、前記関数を用いて、該被験者がどの病期の糖尿病網膜症であるかの確率を算出することを含む
糖尿病網膜症の病期の判定を支援する方法。
〔２〕 前記被験者の眼底写真の画像データと医師が下した前記被験者の糖尿病網膜症の病期診断データとを前記データ群に加えて、新たなデータ群を得、該新たなデータ群に基いて、機械学習アルゴリズムを用いて、前記関数を更新することを含む、〔１〕に記載の糖尿病網膜症の病期の判定を支援する方法。
[1] Diagnosis of diabetic retinopathy and its corresponding fundus photograph using a machine learning algorithm on the basis of a data group including stage diagnosis data of diabetic retinopathy and image data of a fundus photograph corresponding thereto Generate a function that maps the image data of
A method of assisting in determining the stage of diabetic retinopathy, which comprises calculating the probability of which stage the subject has diabetic retinopathy from the image data of the fundus photograph of the subject using the function.
[2] Image data of a fundus photograph of the subject and staging data of diabetic retinopathy of the subject provided by a doctor are added to the data group to obtain a new data group, and based on the new data group A method for assisting in determining the stage of diabetic retinopathy according to [1], including updating the function using a machine learning algorithm.
〔３〕 糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを写像する関数を有し、該関数を用いて被験者の眼底写真の画像データから該被験者がどの病期の糖尿病網膜症であるかの確率を算出する判定手段、
該判定手段に被験者の眼底写真の画像データを入力する手段、および
前記判定手段が算出した前記確率を出力する手段
を具える糖尿病網膜症の病期判定支援システム。
〔４〕 前記関数が、糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを含むデータ群に基いて、機械学習アルゴリズムを用いて、生成させたものである、〔３〕に記載の糖尿病網膜症の病期判定支援システム。
〔５〕 前記被験者の眼底写真の画像データと医師が下した前記被験者の糖尿病網膜症の病期診断データとに基いて、前記関数を更新する手段をさらに具える〔３〕または〔４〕に記載の糖尿病網膜症の病期判定支援システム。
[3] A function for mapping the stage diagnosis data of diabetic retinopathy and the image data of the fundus photograph corresponding thereto, and using the function, the stage of diabetes of the subject from the image data of the fundus photograph of the subject Determination means for calculating the probability of retinopathy,
A stage determination support system for diabetic retinopathy, comprising: a unit for inputting image data of a fundus photograph of a subject to the determination unit; and a unit for outputting the probability calculated by the determination unit.
[4] The function is generated by using a machine learning algorithm based on a data group including stage diagnosis data of diabetic retinopathy and image data of a fundus photograph corresponding thereto, [3] The stage determination support system for diabetic retinopathy according to.
[5] Further comprising means for updating the function based on the image data of the fundus photograph of the subject and the stage diagnosis data of diabetic retinopathy of the subject provided by the doctor. [3] or [4] The staging determination support system for diabetic retinopathy as described.
〔６〕 糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを写像する関数が、
被験者の眼底写真の画像データのコントラストを画面全体一律に調整し、
その調整で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定し、
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合、１番目に明るい部分と２番目に明るい部分との中心間距離を算出し、
算出された中心間距離が基準距離の範囲内であるか否かを判定することを含むものである、〔３〕に記載の糖尿病網膜症の病期判定支援システム。
[6] A function for mapping the stage diagnosis data of diabetic retinopathy and the corresponding image data of the fundus photograph is
Adjust the contrast of the image data of the subject's fundus photograph uniformly over the entire screen,
With that adjustment, when the diameter of the first bright part is set within the range of the reference diameter, it is determined whether the diameter of the second bright part is within the range of the reference diameter,
When it is determined that the diameter of the second brightest part is within the range of the reference diameter, the center distance between the first brightest part and the second brightest part is calculated,
The stage determination support system for diabetic retinopathy according to [3], including determining whether or not the calculated center-to-center distance is within a range of a reference distance.
〔７〕 基準直径が画角で１〜３度であり、基準距離が画角で５〜２５度である、〔６〕に記載の糖尿病網膜症の病期判定支援システム。
〔８〕 基準直径が画角で１．５〜２．５度であり、基準距離が画角で６〜２４度である、〔６〕に記載の糖尿病網膜症の病期判定支援システム。
〔９〕 基準直径が画角で１．８〜２．３度であり、基準距離が画角で６．１〜２３．７度である、〔６〕に記載の糖尿病網膜症の病期判定支援システム。
[7] The stage determination support system for diabetic retinopathy according to [6], wherein the reference diameter is 1 to 3 degrees in angle of view and the reference distance is 5 to 25 degrees in angle of view.
[8] The staging determination support system for diabetic retinopathy according to [6], wherein the reference diameter is 1.5 to 2.5 degrees in angle of view and the reference distance is 6 to 24 degrees in angle of view.
[9] The stage determination of diabetic retinopathy according to [6], wherein the reference diameter is 1.8 to 2.3 degrees in angle of view and the reference distance is 6.1 to 23.7 degrees in angle of view. Support system.
〔１０〕 被験者の眼底を眼底カメラにて撮影して被験者の眼底写真の画像データを得、
該被験者の眼底写真の画像データのコントラストを画面全体一律に調整し、
その調整で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定し、
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合、１番目に明るい部分と２番目に明るい部分との中心間距離を算出し、
算出された中心間距離が基準距離の範囲内であるか否かを判定することを含む糖尿病網膜症の病期の判定を支援する方法。
[10] A subject's fundus is photographed by a fundus camera to obtain image data of a subject's fundus photograph,
The contrast of the image data of the fundus photograph of the subject is adjusted uniformly over the entire screen,
With that adjustment, when the diameter of the first bright part is set within the range of the reference diameter, it is determined whether the diameter of the second bright part is within the range of the reference diameter,
When it is determined that the diameter of the second brightest part is within the range of the reference diameter, the center distance between the first brightest part and the second brightest part is calculated,
A method for supporting the determination of the stage of diabetic retinopathy, which comprises determining whether the calculated center-to-center distance is within a range of a reference distance.
〔１１〕 基準直径が画角で１〜３度であり、基準距離が画角で５〜２５度である、〔１０〕に記載の糖尿病網膜症の病期の判定を支援する方法。
〔１２〕 基準直径が画角で１．５〜２．５度であり、基準距離が画角で６〜２４度である、〔１０〕に記載の糖尿病網膜症の病期の判定を支援する方法。
〔１３〕 基準直径が画角で１．８〜２．３度であり、基準距離が画角で６．１〜２３．７度である、〔１０〕に記載の糖尿病網膜症の病期の判定を支援する方法。
[11] The method for supporting the determination of the stage of diabetic retinopathy according to [10], wherein the reference diameter is 1 to 3 degrees in angle of view and the reference distance is 5 to 25 degrees in angle of view.
[12] The determination of the stage of diabetic retinopathy according to [10], wherein the reference diameter is 1.5 to 2.5 degrees in the angle of view and the reference distance is 6 to 24 degrees in the angle of view. Method.
[13] The stage of diabetic retinopathy according to [10], wherein the reference diameter is 1.8 to 2.3 degrees in angle of view and the reference distance is 6.1 to 23.7 degrees in angle of view. How to help the decision.
〔１４〕 中心間距離が画角で９．７度以下である場合、増殖糖尿病網膜症である確率が高く、中心間距離が画角で９．７度超過である場合、前増殖糖尿病網膜症である確率が高いとすることをさらに含む、〔１１〕〜〔１３〕のいずれかひとつに記載の糖尿病網膜症の病期の判定を支援する方法。
〔１５〕 中心間距離が画角で１３．５度以下である場合、増殖糖尿病網膜症である確率が高く、中心間距離が画角で１３．５度超過である場合、前増殖糖尿病網膜症である確率が高いとすることをさらに含む、〔１１〕〜〔１３〕のいずれかひとつに記載の糖尿病網膜症の病期の判定を支援する方法。
〔１６〕 中心間距離が画角で１６．９度以下である場合、増殖糖尿病網膜症である確率が高く、中心間距離が画角で１６．９度超過である場合、前増殖糖尿病網膜症である確率が高いとすることをさらに含む、〔１１〕〜〔１３〕のいずれかひとつに記載の糖尿病網膜症の病期の判定を支援する方法。
[14] If the center-to-center distance is 9.7 degrees or less in the angle of view, there is a high probability of proliferative diabetic retinopathy, and if the center-to-center distance is more than 9.7 degrees in the angle of view, preproliferative diabetic retinopathy. The method for supporting the determination of the stage of diabetic retinopathy according to any one of [11] to [13], further comprising:
[15] If the center-to-center distance is 13.5 degrees or less at the angle of view, there is a high probability of proliferative diabetic retinopathy, and if the center-to-center distance is more than 13.5 degrees at the angle of view, preproliferative diabetic retinopathy. The method for supporting the determination of the stage of diabetic retinopathy according to any one of [11] to [13], further comprising:
[16] If the center-to-center distance is 16.9 degrees or less in the angle of view, there is a high probability of proliferative diabetic retinopathy, and if the center-to-center distance is more than 16.9 degrees in the angle of view, preproliferative diabetic retinopathy. The method for supporting the determination of the stage of diabetic retinopathy according to any one of [11] to [13], further comprising:
〔１７〕 被験者の眼底を撮影して眼底写真の画像データを得るための眼底カメラ、
被験者の眼底写真の画像データのコントラストを画面全体一律に調整する手段、
調整する手段で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定する手段、
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合に、１番目に明るい部分と２番目に明るい部分との中心間距離を算出する手段、および
算出された中心間距離が基準距離の範囲内にあるか否かを判定する手段
を有する糖尿病網膜症の病期判定支援システム。
[17] A fundus camera for photographing the fundus of the subject to obtain image data of a fundus photograph,
A means for uniformly adjusting the contrast of the image data of the fundus photograph of the test subject on the entire screen,
Means for determining whether or not the diameter of the second brightest portion is within the range of the reference diameter when the diameter of the first brightest portion is within the range of the reference diameter by the adjusting means,
A means for calculating the center distance between the first bright portion and the second bright portion when it is determined that the diameter of the second bright portion is within the range of the reference diameter, and the calculated center distance. A stage determination support system for diabetic retinopathy having means for determining whether or not is within the range of the reference distance.
本発明によれば、撮影された被験者の眼底写真から糖尿病網膜症の病期を簡易的且つ定量的に評価することができる。本発明の一実施形態は、高い明度で且つ略円形に描出される部分に着目するので、低い明度で細いスジで描出される血管部分よりも、見落としが低減できる。 According to the present invention, the stage of diabetic retinopathy can be easily and quantitatively evaluated from the photographed fundus photograph of the subject. Since one embodiment of the present invention focuses on a portion that is drawn in a substantially circular shape with high brightness, oversight can be reduced compared to a blood vessel portion that is drawn with a thin stripe with low brightness.
本発明の糖尿病網膜症の病期判定支援システムは、糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを写像する関数を有し、該関数を用いて被験者の眼底写真の画像データから該被験者がどの病期の糖尿病網膜症であるかの確率を算出する判定手段、該判定手段に被験者の眼底写真の画像データを入力する手段、および前記判定手段が算出した前記確率を出力する手段を具えるものである。 The stage determination support system for diabetic retinopathy of the present invention has a function for mapping the stage diagnosis data of diabetic retinopathy and the image data of the fundus photograph corresponding thereto, and using the function of the fundus photograph of the subject. Judgment means for calculating the probability of which stage the subject has diabetic retinopathy from the image data, means for inputting image data of the fundus photograph of the subject to the judgment means, and the probability calculated by the judgment means It is equipped with a means for outputting.
本発明の糖尿病網膜症の病期の判定を支援する方法は、被験者の眼底写真の画像データから、糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを写像する関数を用いて、該被験者がどの病期の糖尿病網膜症であるかの確率を算出することを含む。 The method for supporting the determination of the stage of diabetic retinopathy of the present invention, from the image data of the fundus photograph of the subject, using the function to map the stage diagnosis data of diabetic retinopathy and the image data of the corresponding fundus photograph. And calculating the probability of which stage the subject has diabetic retinopathy.
前記関数は、糖尿病網膜症の病期診断データとそれに対応する眼底写真の画像データとを含むデータ群に基いて、機械学習アルゴリズムを用いて、生成させることができる。機械学習としては、教師あり学習、半教師あり学習、教師なし学習、深層学習などが挙げられる。本発明においては、機械学習アルゴリズムとして、教師あり学習アルゴリズム若しくは半教師あり学習アルゴリズムまたは深層学習アルゴリズムが好ましく用いられる。深層学習アルゴリズムによると、関数を特徴づける変数が自動的に設定され、計算される。 The function can be generated using a machine learning algorithm based on a data group including stage diagnosis data of diabetic retinopathy and image data of a fundus photograph corresponding thereto. Examples of machine learning include supervised learning, semi-supervised learning, unsupervised learning, and deep learning. In the present invention, a supervised learning algorithm, a semi-supervised learning algorithm, or a deep learning algorithm is preferably used as the machine learning algorithm. According to the deep learning algorithm, variables characterizing the function are automatically set and calculated.
前記被験者の眼底写真の画像データと医師が下した前記被験者の糖尿病網膜症の病期診断データとを前記データ群に加えて、新たなデータ群を得、該新たなデータ群に基いて、機械学習アルゴリズムを用いて、前記関数を更新することができる。関数の更新は、例えば、データマイニングなどで用いられる手法により行うことができる。該手法の具体例としては、分類、回帰分析、予測、勾配ブースティング、頻出パターン抽出などが挙げられる。 Image data of the fundus photograph of the subject and staging data of diabetic retinopathy of the subject made by a doctor are added to the data group to obtain a new data group, and based on the new data group, a machine A learning algorithm can be used to update the function. The function can be updated by, for example, a method used in data mining or the like. Specific examples of the method include classification, regression analysis, prediction, gradient boosting, and frequent pattern extraction.
本発明に用いられる関数は、該関数を特徴づける変数を手動で設定し計算したものであってもよい。例えば、本発明に用いられる関数は、被験者の眼底写真の画像データのコントラストを画面全体一律に調整し、 その調整で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定し、 ２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合、１番目に明るい部分と２番目に明るい部分との中心間距離を算出し、算出された中心間距離が基準距離の範囲内であるか否かを判定することを含むものであってもよい。本発明の一実施形態によると、機械学習アルゴリズムによって生成した関数も上記のような機能を少なくとも有する関数であった。 The function used in the present invention may be one in which variables that characterize the function are manually set and calculated. For example, the function used in the present invention is to adjust the contrast of the image data of the fundus photograph of the subject uniformly over the entire screen, and the adjustment is performed so that the diameter of the brightest part is within the range of the reference diameter. Then, it is determined whether the diameter of the second brightest part is within the range of the reference diameter. If it is determined that the diameter of the second brightest part is within the range of the reference diameter, the first brightest part And calculating the center-to-center distance between the second brightest part and determining whether or not the calculated center-to-center distance is within the range of the reference distance. According to the embodiment of the present invention, the function generated by the machine learning algorithm is also a function having at least the above-mentioned function.
被験者の眼底写真の画像データは、黄斑部の中央が画像中心に、視神経乳頭が３時または９時の方向になるように位置合わせをして撮影された被験者の眼底写真の画像データであることが好ましい。右目と左目とでは、黄班部と視神経乳頭との位置関係が逆になるので、画像データ処理を簡便にするために、黄班部と視神経乳頭の位置関係が同じになるように、右目または左目のいずれか一方の画像データは反転させて使用することができる。また、後述する明るい部分の判定において黄班部および視神経乳頭の明部が判定のノイズとならないようにするために、正常者の眼底を撮影して得た眼底写真の標準画像データを用意して、被験者の眼底写真の画像データから眼底写真の標準画像データを差し引いて得られる差分画像データを用いてもよい。 The image data of the fundus photograph of the subject should be the image data of the fundus photograph of the subject that was taken by aligning the center of the macula with the image center and the optic disc at 3 o'clock or 9 o'clock. Is preferred. Since the positional relationship between the macula and the optic disc is reversed between the right eye and the left eye, in order to simplify the image data processing, the positional relationship between the macula and the optic disc should be the same. The image data of either one of the left eyes can be inverted and used. Further, in order to prevent the macular part and the bright part of the optic disc from becoming noise in the judgment of the bright part described later, prepare standard image data of the fundus photograph obtained by photographing the fundus of a normal person. The difference image data obtained by subtracting the standard image data of the fundus photograph from the image data of the fundus photograph of the subject may be used.
図６は正常者の眼底写真の画像データの一例を示す図である。正常者の画像データは、コントラストを調整しても、画面全体の明度が一律変化するだけで、背景部分に比べて輝点のような明るい部分を観測することがない。一方、糖尿病網膜症患者の眼底写真には、軟性白斑（図１）、硬性白斑（図２）、光凝固班（図３）、増殖膜（図４）、内境界膜肥厚（図５）などによる、背景部分に比べて明るい部分が観測される。明るい部分の大きさや明度は、被験者毎にまたは撮影環境毎に異なる。
そこで、本発明では、被験者の眼底写真の画像データのコントラストを画面全体一律に調整し、 その調整で、１番目に明るい部分L1の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分L2の直径が基準直径の範囲内にあるか否かを第一の判定基準とした。基準直径は、糖尿病網膜症の判定の感度および特異度が高くなるように、例えば、ＲＯＣ解析などを用いて、適宜設定することができ、例えば、画角で、好ましくは１〜３度、より好ましくは１．５〜２．５度、さらに好ましくは１．８〜２．３度に設定することができる。
FIG. 6 is a diagram showing an example of image data of a fundus photograph of a normal person. In the image data of a normal person, even if the contrast is adjusted, the brightness of the entire screen changes uniformly, and a bright portion such as a bright spot is not observed compared to the background portion. On the other hand, in a fundus photograph of a diabetic retinopathy patient, soft vitiligo (Fig. 1), hard vitiligo (Fig. 2), photocoagulation plaque (Fig. 3), proliferative membrane (Fig. 4), internal limiting membrane thickening (Fig. 5), etc. The bright part is observed compared to the background part. The size and brightness of the bright portion differ depending on the subject or the shooting environment.
Therefore, in the present invention, the contrast of the image data of the fundus photograph of the subject is uniformly adjusted over the entire screen, and when the adjustment is performed so that the diameter of the brightest part L1 is within the range of the reference diameter, The first criterion was whether the diameter of the second brightest part L2 was within the range of the reference diameter. The reference diameter can be appropriately set, for example, using ROC analysis or the like so that the sensitivity and specificity of the determination of diabetic retinopathy are high. For example, the angle of view is preferably 1 to 3 degrees. It can be set to preferably 1.5 to 2.5 degrees, and more preferably 1.8 to 2.3 degrees.
上記のコントラスト調整で、２番目に明るい部分の直径が基準直径の範囲内にある場合は、増殖糖尿病網膜症または前増殖糖尿病網膜症である確率が高く、２番目に明るい部分の直径が基準直径の範囲内にない場合は、増殖糖尿病網膜症でも前増殖糖尿病網膜症でもない確率が高いと判定できる。 In the above contrast adjustment, if the diameter of the second brightest part is within the reference diameter range, the probability of proliferative diabetic retinopathy or preproliferative diabetic retinopathy is high, and the diameter of the second brightest part is the reference diameter. If it is not within the range, it can be determined that there is a high probability that neither proliferative diabetic retinopathy nor preproliferative diabetic retinopathy is present.
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合、１番目に明るい部分と２番目に明るい部分との中心間距離を算出する。そして、本発明では、算出された中心間距離が基準距離の範囲内であるか否かを第二の判定基準とした。基準距離は、糖尿病網膜症の判定の感度および特異度が高くなるように、例えば、ＲＯＣ解析などを用いて、適宜設定することができ、例えば、画角で、好ましくは５〜２５度、より好ましくは６〜２４度、さらに好ましくは６．１〜２３．７度に設定することができる。算出された中心間距離が基準距離の範囲内である場合、前記被験者が増殖糖尿病網膜症または前増殖糖尿病網膜症である確率が高く、算出された中心間距離が基準距離の範囲内でない場合、前記被験者が前増殖糖尿病網膜症でも増殖糖尿病網膜症でもない確率が高いとすることを含むものであってもよい。 When it is determined that the diameter of the second brightest portion is within the range of the reference diameter, the center-to-center distance between the first brightest portion and the second brightest portion is calculated. Then, in the present invention, whether or not the calculated center-to-center distance is within the range of the reference distance is the second determination criterion. The reference distance can be appropriately set, for example, by using ROC analysis or the like so that the sensitivity and specificity of the determination of diabetic retinopathy are high. For example, the angle of view is preferably 5 to 25 degrees, and more preferably It can be set to preferably 6 to 24 degrees, and more preferably 6.1 to 23.7 degrees. If the calculated center distance is within the range of the reference distance, the subject is likely to have proliferative diabetic retinopathy or preproliferative diabetic retinopathy, if the calculated center distance is not within the range of the reference distance, It may include that the subject has a high probability of not having preproliferative diabetic retinopathy or proliferative diabetic retinopathy.
算出された中心間距離が基準距離の範囲内のうちの短い側の範囲内である場合、前記被験者が増殖糖尿病網膜症である確率が高いとすることができ、算出された中心間距離が基準距離の範囲内のうちで長い側の領域である場合、前増殖糖尿病網膜症である確率が高いとすることができる。
基準距離の範囲内において短い側と長い側とを区別する閾値は、糖尿病網膜症の判定の感度および特異度が高くなるように、例えば、ＲＯＣ解析などを用いて、適宜設定することができる。該閾値は、例えば、９．７度〜１６．９度の範囲内のある一つの値とすることができる。
When the calculated center-to-center distance is within the range on the shorter side of the range of the reference distance, it can be considered that the subject has a high probability of having proliferative diabetic retinopathy, and the calculated center-to-center distance is the reference. If the region is on the long side within the range of the distance, it can be determined that the probability of preproliferative diabetic retinopathy is high.
The threshold for distinguishing the short side and the long side within the range of the reference distance can be appropriately set by using, for example, ROC analysis so that the sensitivity and specificity of the determination of diabetic retinopathy are high. The threshold can be, for example, a certain value within the range of 9.7 degrees to 16.9 degrees.
以下に実施例を示して本発明をより詳細に説明する。なお、本発明は本実施例によって限定されない。 Hereinafter, the present invention will be described in more detail with reference to examples. The present invention is not limited to this embodiment.
実施例１
定期健康診断と同じ条件で撮影された撮影画角４５度の眼底写真約１００００枚とそれに対応する精密検査によって診断された糖尿病網膜症病期分類とを対比させた。
眼底写真に、直径が画角で２．０３（対数正規分布で９５％信頼区間１．８５−２．２３、９９％信頼区間１．７９−２．３０）度である少なくとも２つの輝点のような明るい部分が観測された場合、その患者は精密検査による診断で前増殖糖尿病網膜症または増殖糖尿病網膜症であった。
該２つの輝点の中心間距離が画角で１８．２０（正規分布で９５％信頼区間１４．００−２２．３９、９９％信頼区間１２．６８−２３．７２）度であった場合、その患者は精密検査による診断で前増殖糖尿病網膜症であり、前記中心間距離が画角で８．２６（対数正規分布で９５％信頼区間６．５４−１０．４４、９９％信頼区間６．０７−１１．２４）度であった場合、その患者は精密検査による診断で増殖糖尿病網膜症であった。
前増殖糖尿病網膜症であるか増殖糖尿病網膜症であるかの閾値は、ＲＯＣ解析から、９．７度と１６．９度の間、好ましくは１３．５度であった。
本発明のシステムまたは方法は、２つの明るい円形状の部分という明確かつ単純な特徴量をもって前増殖糖尿病網膜症を感度９５％、特異度１００％で判定できた。
２つの円の中心間距離の閾値を、画角で１３．５度に設定した場合、前増殖糖尿病網膜症に対する増殖糖尿病網膜症の感度が９０％、特異度が７８％であった。
２つの円の中心間距離の閾値を、画角で９．７度に設定した場合、前増殖糖尿病網膜症に対する増殖糖尿病網膜症の感度が７５％、特異度が８３％であった。
２つの円の中心間距離の閾値を、画角で１６．９度に設定した場合、前増殖糖尿病網膜症に対する増殖糖尿病網膜症の感度が９５％、特異度が５０％であった。
このように、本発明のシステムまたは方法は、病期判定支援に有用である。
Example 1
Approximately 10000 pieces of fundus photographs taken at the angle of view of 45 degrees taken under the same conditions as the regular health examination were compared with the diabetic retinopathy staging classified by the corresponding detailed examination.
A fundus photograph shows at least two bright spots having a diameter of 2.03 (angle of log normal distribution: 95% confidence interval 1.85-2.23, 99% confidence interval 1.79-2.30). If such a bright spot was observed, the patient had preproliferative diabetic retinopathy or proliferative diabetic retinopathy as diagnosed by work-up.
When the distance between the centers of the two bright spots is 18.20 in view angle (95% confidence interval 14.00-22.39 in normal distribution, 99% confidence interval 12.68-23.72), The patient had preproliferative diabetic retinopathy as determined by close examination, and the center-to-center distance was 8.26 in angle of view (95% confidence interval 6.54-10.44 in lognormal distribution, 99% confidence interval 6. 07-11.24), the patient had proliferative diabetic retinopathy as diagnosed by close examination.
The threshold for preproliferative diabetic retinopathy or proliferative diabetic retinopathy was between 9.7 and 16.9 degrees, preferably 13.5 degrees, from ROC analysis.
The system or method of the present invention was able to determine preproliferative diabetic retinopathy with a sensitivity of 95% and a specificity of 100% with clear and simple characteristic amounts of two bright circular portions.
When the threshold value of the distance between the centers of the two circles was set to 13.5 degrees in the angle of view, the sensitivity of proliferative diabetic retinopathy to preproliferative diabetic retinopathy was 90% and the specificity was 78%.
When the threshold value of the distance between the centers of the two circles was set to 9.7 degrees in the angle of view, the sensitivity of proliferative diabetic retinopathy to preproliferative diabetic retinopathy was 75% and the specificity was 83%.
When the threshold value of the distance between the centers of the two circles was set to 16.9 degrees in the angle of view, the sensitivity of proliferative diabetic retinopathy to preproliferative diabetic retinopathy was 95% and the specificity was 50%.
Thus, the system or method of the present invention is useful for staging assistance.
比較例１
眼底写真において硬性白斑または軟性白斑があるということで単純糖尿病網膜症であると判定された７例の患者は、精密検査による診断で前増殖糖尿病網膜症または増殖糖尿病網膜症であることが後に判明した。
Comparative Example 1
Seven patients who were determined to have simple diabetic retinopathy due to hard or soft vitiligo on fundus photographs were later found to have preproliferative diabetic retinopathy or proliferative diabetic retinopathy by physical examination. did.
実施例２
比較例１で使用した７例の患者の眼底写真の画像データのコントラストを調整したところ、本発明の規定する輝点のような明るい部分が少なくとも２つ観測された。本発明のシステムまたは方法によると前増殖糖尿病網膜症または増殖糖尿病網膜症である確率が高いとの判定ができた。
Example 2
When the contrast of the image data of the fundus photographs of the seven patients used in Comparative Example 1 was adjusted, at least two bright parts such as bright spots defined by the present invention were observed. According to the system or method of the present invention, it was possible to determine that the probability of preproliferative diabetic retinopathy or proliferative diabetic retinopathy is high.
比較例２
内境界膜肥厚が観測されたが糖尿病網膜症であると判定されなかった２例の患者は、精密検査による診断で前増殖糖尿病網膜症または増殖糖尿病網膜症であることが後に判明した。
Comparative example 2
Two patients who had an internal limiting membrane thickening but were not determined to have diabetic retinopathy were later found to have preproliferative diabetic retinopathy or proliferative diabetic retinopathy by physical examination.
実施例３
比較例２で使用した２例の患者の眼底写真の画像データのコントラストを調整したところ、本発明の規定する輝点のような明るい部分が少なくとも２つ観測された。本発明のシステムまたは方法によると前増殖糖尿病網膜症または増殖糖尿病網膜症である確率が高いとの判定ができた。
Example 3
When the contrast of the image data of the fundus photographs of the two patients used in Comparative Example 2 was adjusted, at least two bright parts such as bright spots defined by the present invention were observed. According to the system or method of the present invention, it was possible to determine that the probability of preproliferative diabetic retinopathy or proliferative diabetic retinopathy is high.
本発明のシステムまたは方法は、それを眼底カメラ又は電子カルテに組み込むことで、定期健康診断、人間ドック、外来患者診断などにおいて、糖尿病網膜症の病期判定をより正確に行うための支援ができる。 The system or method of the present invention can be incorporated into a fundus camera or an electronic medical chart to assist in more accurately determining the stage of diabetic retinopathy in a periodic health checkup, medical checkup, outpatient diagnosis, and the like.
Claims (10)
新たな眼底写真の画像データから、前記関数を用いて、それに対応する糖尿病網膜症の病期診断データに写像することを含み、
前記の画像データは、いずれも、１番目に明るい部分の直径が基準直径の範囲内になるように、画面全体一律にコントラストが調整されたものである、
糖尿病網膜症の病期の判定を支援する方法。 Based on the data group including the stage diagnosis data of diabetic retinopathy and the image data of the fundus photograph corresponding thereto, using the machine learning algorithm, the stage diagnosis data of diabetic retinopathy and the image data of the corresponding fundus photograph Generate a function that maps and,
From the new fundus photograph of the image data, by using the function, see contains that maps to staging data corresponding diabetic retinopathy thereto,
In each of the above image data, the contrast is uniformly adjusted over the entire screen so that the diameter of the brightest part is within the range of the reference diameter.
A method to help determine the stage of diabetic retinopathy.
該判定手段に被験者の眼底写真の画像データを入力する手段、および
前記判定手段が算出した前記確率を出力する手段
を具え、
前記の画像データは、いずれも、１番目に明るい部分の直径が基準直径の範囲内になるように、画面全体一律にコントラストが調整されたものである、
糖尿病網膜症の病期判定支援システム。 Having a function for mapping the stage diagnosis data of diabetic retinopathy and the image data of the fundus photograph corresponding to it, which stage the diabetic retinopathy of the subject is from the image data of the fundus photograph of the subject using the function. Determination means for calculating the probability of whether,
The determining means comprises means for inputting image data of a fundus photograph of the subject, and means for outputting the probability calculated by the determining means ,
In each of the above image data, the contrast is uniformly adjusted over the entire screen so that the diameter of the brightest part is within the range of the reference diameter.
Diagnosis system for diabetic retinopathy.
算出された中心間距離が基準距離の範囲内であるか否かを判定することを含む、請求項３〜６のいずれかひとつに記載の糖尿病網膜症の病期判定支援システム。The stage determination support system for diabetic retinopathy according to any one of claims 3 to 6, including determining whether the calculated center-to-center distance is within a range of a reference distance.
該判定手段に被験者の眼底写真の画像データを入力する手段、および
前記判定手段が算出した前記確率を出力する手段
を具え、
前記関数が、
被験者の眼底写真の画像データのコントラストを画面全体一律に調整し、
その調整で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定し、
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合、１番目に明るい部分と２番目に明るい部分との中心間距離を算出し、
算出された中心間距離が基準距離の範囲内であるか否かを判定することを含むものである、糖尿病網膜症の病期判定支援システム。 Having a function for mapping the stage diagnosis data of diabetic retinopathy and the image data of the fundus photograph corresponding thereto, the diabetic retinopathy of which stage the subject is from the image data of the fundus photograph of the subject using the function. Determination means for calculating the probability of whether,
Means for inputting image data of a fundus photograph of the subject to the determination means, and
Means for outputting the probability calculated by the judging means
With
The function is
Adjust the contrast of the image data of the subject's fundus photograph uniformly over the entire screen,
With that adjustment, when the diameter of the first bright part is set within the range of the reference diameter, it is determined whether the diameter of the second bright part is within the range of the reference diameter,
When it is determined that the diameter of the second brightest part is within the range of the reference diameter, the center distance between the first brightest part and the second brightest part is calculated,
It is intended to include the calculated distance between the centers to determine whether it is within range of the reference distance, staging support system diabetes retinopathy.
該被験者の眼底写真の画像データのコントラストを画面全体一律に調整し、
その調整で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定し、
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合、１番目に明るい部分と２番目に明るい部分との中心間距離を算出し、
算出された中心間距離が基準距離の範囲内であるか否かを判定することを含む糖尿病網膜症の病期の判定を支援する方法。 Obtain the image data of the subject's fundus photograph by photographing the subject's fundus with a fundus camera,
The contrast of the image data of the fundus photograph of the subject is adjusted uniformly over the entire screen,
With the adjustment, when the diameter of the first bright part is set to be within the range of the reference diameter, it is determined whether the diameter of the second bright part is within the range of the reference diameter,
When it is determined that the diameter of the second brightest part is within the range of the reference diameter, the center distance between the first brightest part and the second brightest part is calculated,
A method for supporting the determination of the stage of diabetic retinopathy, which comprises determining whether the calculated center-to-center distance is within a range of a reference distance.
被験者の眼底写真の画像データのコントラストを画面全体一律に調整する手段、
調整する手段で、１番目に明るい部分の直径が基準直径の範囲内になるようにしたときに、２番目に明るい部分の直径が基準直径の範囲内にあるか否かを判定する手段、
２番目に明るい部分の直径が基準直径の範囲内にあると判定された場合に、１番目に明るい部分と２番目に明るい部分との中心間距離を算出する手段、および
算出された中心間距離が基準距離の範囲内にあるか否かを判定する手段
を有する糖尿病網膜症の病期判定支援システム。 A fundus camera for photographing the fundus of the subject to obtain image data of a fundus photograph,
A means for uniformly adjusting the contrast of the image data of the fundus photograph of the test subject on the entire screen,
Means for determining whether or not the diameter of the second brightest portion is within the range of the reference diameter when the diameter of the first brightest portion is within the range of the reference diameter by the adjusting means,
A means for calculating the center distance between the first bright portion and the second bright portion when it is determined that the diameter of the second bright portion is within the range of the reference diameter, and the calculated center distance. A stage determination support system for diabetic retinopathy having means for determining whether or not is within the range of the reference distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016161579A JP6745496B2 (en) | 2016-08-19 | 2016-08-19 | Diabetic retinopathy stage determination support system and method for supporting stage determination of diabetic retinopathy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016161579A JP6745496B2 (en) | 2016-08-19 | 2016-08-19 | Diabetic retinopathy stage determination support system and method for supporting stage determination of diabetic retinopathy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JP2018027273A JP2018027273A (en) | 2018-02-22 |
| JP6745496B2 true JP6745496B2 (en) | 2020-08-26 |
Family
ID=61248007
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP2016161579A Active JP6745496B2 (en) | 2016-08-19 | 2016-08-19 | Diabetic retinopathy stage determination support system and method for supporting stage determination of diabetic retinopathy |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JP6745496B2 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220079430A1 (en) * | 2017-05-04 | 2022-03-17 | Shenzhen Sibionics Technology Co., Ltd. | System for recognizing diabetic retinopathy |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP7196908B2 (en) * | 2018-04-27 | 2022-12-27 | 株式会社ニデック | Ophthalmic image processing device and ophthalmic image processing program |
| CN111292286B (en) * | 2018-11-21 | 2023-07-11 | 福州依影健康科技有限公司 | Analysis method and system for change of characteristic data of sugar mesh bottom and storage device |
| KR102330519B1 (en) * | 2018-12-03 | 2021-11-24 | 서울대학교병원 | Apparatus for predicting intracranial aneurysm using retinal fundus image and method for providing intracranial aneurysm prediction results using the same |
| KR102058883B1 (en) * | 2019-04-11 | 2019-12-24 | 주식회사 홍복 | Method of analyzing iris image and retina image for diagnosing diabetes and pre-symptom in artificial intelligence |
| KR102313143B1 (en) * | 2019-07-23 | 2021-10-18 | 단국대학교 산학협력단 | Diabetic retinopathy detection and severity classification apparatus Based on Deep Learning and method thereof |
| TWI702615B (en) * | 2019-07-26 | 2020-08-21 | 長佳智能股份有限公司 | Retinopathy assessment model establishment method and system |
| CA3178829C (en) | 2020-05-14 | 2024-02-20 | Peter Robert Wakeford | Retinal image processing |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005508215A (en) * | 2001-08-30 | 2005-03-31 | フィラデルフィア オフサルミック イメージング システムズ | System and method for screening patients with diabetic retinopathy |
| WO2008062528A1 (en) * | 2006-11-24 | 2008-05-29 | Nidek Co., Ltd. | Fundus image analyzer |
| JP5726081B2 (en) * | 2009-09-16 | 2015-05-27 | 株式会社日立メディコ | Ultrasonic diagnostic apparatus and elasticity image classification program |
| CN103458772B (en) * | 2011-04-07 | 2017-10-31 | 香港中文大学 | Retinal images analysis method and device |
| US20140314288A1 (en) * | 2013-04-17 | 2014-10-23 | Keshab K. Parhi | Method and apparatus to detect lesions of diabetic retinopathy in fundus images |
-
2016
- 2016-08-19 JP JP2016161579A patent/JP6745496B2/en active Active
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220079430A1 (en) * | 2017-05-04 | 2022-03-17 | Shenzhen Sibionics Technology Co., Ltd. | System for recognizing diabetic retinopathy |
| US11666210B2 (en) * | 2017-05-04 | 2023-06-06 | Shenzhen Sibionics Technology Co., Ltd. | System for recognizing diabetic retinopathy |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2018027273A (en) | 2018-02-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6745496B2 (en) | Diabetic retinopathy stage determination support system and method for supporting stage determination of diabetic retinopathy | |
| US10441160B2 (en) | Method and system for classifying optic nerve head | |
| Patton et al. | Retinal image analysis: concepts, applications and potential | |
| Witmer et al. | Wide-field imaging of the retina | |
| Haleem et al. | Automatic extraction of retinal features from colour retinal images for glaucoma diagnosis: a review | |
| Markowitz et al. | Microperimetry and clinical practice: an evidence-based review | |
| US20190313903A1 (en) | System and Method for Medical Condition Diagnosis, Treatment and Prognosis Determination | |
| Hood et al. | A test of a linear model of glaucomatous structure–function loss reveals sources of variability in retinal nerve fiber and visual field measurements | |
| Ahmad Fadzil et al. | Analysis of retinal fundus images for grading of diabetic retinopathy severity | |
| Gerendas et al. | Computational image analysis for prognosis determination in DME | |
| US20170156582A1 (en) | Automated clinical evaluation of the eye | |
| Wallace | Computer-assisted quantification of vascular tortuosity in retinopathy of prematurity (an American Ophthalmological Society thesis) | |
| Hood et al. | Evaluation of a one-page report to aid in detecting glaucomatous damage | |
| Roh et al. | Microperimetry in age-related macular degeneration: association with macular morphology assessed by optical coherence tomography | |
| Bayer et al. | Artifacts and anatomic variations in optical coherence tomography | |
| TW202221637A (en) | Data storage system and data storage method | |
| Murray et al. | Comparison of saccadic vector optokinetic perimetry and standard automated perimetry in glaucoma. Part I: threshold values and repeatability | |
| Kang et al. | Measuring ocular torsion and its variations using different nonmydriatic fundus photographic methods | |
| Attiku et al. | Current status and future possibilities of retinal imaging in diabetic retinopathy care applicable to low-and medium-income countries | |
| Giancardo | Automated fundus images analysis techniques to screen retinal diseases in diabetic patients | |
| Islam et al. | A deep-learning approach for automated OCT en-face retinal vessel segmentation in cases of optic disc swelling using multiple en-face images as input | |
| Seely et al. | Auto-processed retinal vessel shadow view images from bedside optical coherence tomography to evaluate plus disease in retinopathy of prematurity | |
| Shariflou et al. | Objective quantification of spontaneous retinal venous pulsations using a novel tablet-based ophthalmoscope | |
| Malik et al. | Retinal disorder as a biomarker for detection of human diseases | |
| Pinos-Velez et al. | Implementation of support tools for the presumptive diagnosis of Glaucoma through identification and processing of medical images of the human eye |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| A621 | Written request for application examination |
Free format text: JAPANESE INTERMEDIATE CODE: A621 Effective date: 20190311 |
|
| A131 | Notification of reasons for refusal |
Free format text: JAPANESE INTERMEDIATE CODE: A131 Effective date: 20200107 |
|
| A977 | Report on retrieval |
Free format text: JAPANESE INTERMEDIATE CODE: A971007 Effective date: 20191227 |
|
| A521 | Request for written amendment filed |
Free format text: JAPANESE INTERMEDIATE CODE: A523 Effective date: 20200306 |
|
| A521 | Request for written amendment filed |
Free format text: JAPANESE INTERMEDIATE CODE: A523 Effective date: 20200331 |
|
| TRDD | Decision of grant or rejection written | ||
| A01 | Written decision to grant a patent or to grant a registration (utility model) |
Free format text: JAPANESE INTERMEDIATE CODE: A01 Effective date: 20200623 |
|
| A61 | First payment of annual fees (during grant procedure) |
Free format text: JAPANESE INTERMEDIATE CODE: A61 Effective date: 20200720 |
|
| R150 | Certificate of patent or registration of utility model |
Ref document number: 6745496 Country of ref document: JP Free format text: JAPANESE INTERMEDIATE CODE: R150 |
|
| S111 | Request for change of ownership or part of ownership |
Free format text: JAPANESE INTERMEDIATE CODE: R313113 |
|
| R350 | Written notification of registration of transfer |
Free format text: JAPANESE INTERMEDIATE CODE: R350 |
|
| R250 | Receipt of annual fees |
Free format text: JAPANESE INTERMEDIATE CODE: R250 |