WO2023284117A1

WO2023284117A1 - Blood testing instrument, and blood testing and recognition system and method

Info

Publication number: WO2023284117A1
Application number: PCT/CN2021/120218
Authority: WO
Inventors: 杨奕; 陈龙飞; 刘彦彤
Original assignee: 武汉大学
Priority date: 2021-07-12
Filing date: 2021-09-24
Publication date: 2023-01-19
Also published as: CN113588521A; CN113588521B

Abstract

A blood testing instrument, and a blood testing and recognition system and method. The method comprises: introducing blood and a hydrogel precursor into a chip chamber (13) of a micro-fluidic chip (1) through a chip inlet portion (11) of the micro-fluidic chip (1), a micro-hydrogel column (14) being provided in the chip chamber (13), and blood cells being solidified in the micro-hydrogel column (14) by means of blue light exposure to form a hydrogel actuator; acquiring a blood cell original image by means of an image acquisition and transmission assembly; pressing a mechanical transmission assembly (2) such that the blood cells in the hydrogel actuator are immobilized; acquiring a blood cell deformed image by means of the image acquisition and transmission assembly; transmitting the blood cell original image and the blood cell deformed image to a recognition device by means of the image acquisition and transmission assembly; and obtaining classification and recognition result information by means of the recognition device. The problems in the prior art that a blood testing apparatus is large in size, high in cost and complex in operation are solved.

Description

A blood tester, a blood test identification system, and an identification method

technical field

The invention belongs to the technical field of medical detection, and more specifically relates to a blood detector, a blood detection recognition system and a recognition method.

Background technique

Existing blood testing or analysis is usually performed in hospitals or laboratories, and the testing equipment used is usually bulky, expensive, and complicated to operate. How to provide a portable blood tester to realize high-precision, low-cost, and easy-to-operate multifunctional blood test is still a challenge in this field.

Contents of the invention

The present invention solves the problems of large volume, high cost and complicated operation of the blood detection device in the prior art by providing a blood detection instrument, a blood detection recognition system and a recognition method.

The invention provides a blood detector, comprising: a microfluidic chip, a mechanical transmission component, an image acquisition and transmission component, and a casing;

The microfluidic chip includes a chip inlet, a chip outlet, a chip chamber and a micro hydrogel column;

The microfluidic chip and the image acquisition transmission assembly are installed inside the housing; the mechanical transmission assembly is installed on the housing and arranged above the micro hydrogel column;

The microfluidic chip is used to provide an imaging platform for blood cells; blood and hydrogel precursors are passed into the chip chamber through the chip inlet, and the micro hydrogel column is located in the chip chamber , curing blood cells in the micro-hydrogel column by exposure to blue light to make a hydrogel actuator;

The force transmission component is used to transmit the force generated by external pressing to the hydrogel actuator, so that the blood cells in the hydrogel actuator brake;

The image acquisition and transmission component is used to acquire the original blood cell image and the blood cell deformation image, and transmit the blood cell original image and the blood cell deformation image to an external identification device.

Preferably, the force transmission component includes: a spring, a button and a washer;

One end of the spring is connected to the gasket, and the spring is arranged above the gasket; the button is connected to the other end of the spring, and the button is arranged above the spring; the pad The lower surface of the sheet is in close contact with the upper surface of the micro hydrogel column.

Preferably, the image acquisition and transmission component includes: a patch-type light source module, an optical lens, an image acquisition module and an image transmission module;

The patch light source module is arranged above the microfluidic chip, the optical lens, the image acquisition module and the image transmission module are all arranged under the microfluidic chip, and the image acquisition The modules are respectively connected with the image transmission module and the optical lens;

The patch-type light source module is used to provide a light source for blood cell imaging; the optical lens is used to adjust the focus of blood cell imaging; the image acquisition module is used to acquire the original image of blood cells and the deformed image of blood cells; The image transmission module is used to transmit the original blood cell image and the deformed blood cell image to an external identification device.

The present invention provides a blood detection and recognition system, including: recognition equipment, and the above-mentioned blood detector;

The recognition device is connected to the blood detector through a data line; the recognition device is used to receive the original blood cell image and the blood cell deformation image, and obtain classification and recognition result information according to the blood cell original image and the blood cell deformation image.

Preferably, the recognition device includes an image preprocessing module, a training optimization module, and a classification and recognition module;

The image preprocessing module is used to preprocess the obtained blood cell original image and blood cell deformation image to obtain an image vector and a parameter vector of blood cells; the image vector includes an original image vector and a deformed image vector, and the parameter vector includes a shape The vector of scientific parameters and the vector of mechanical parameters;

The training optimization module is used to train and optimize the pre-built blood detection classification model to obtain a trained blood detection classification model;

The classification and identification module is used to preprocess the original image of the blood cell to be classified and identified and the corresponding deformed image of the blood cell, and then input it into the trained blood detection and classification model to obtain the classification and identification result information.

Preferably, the blood detection classification model is constructed based on a deep convolutional neural network, and the blood detection classification model includes six convolutional layers and three fully connected layers; the three fully connected layers contain 40, 64 and 20 vectors respectively , where the first fully connected layer contains 32 image vectors and 8 parameter vectors.

Preferably, the preprocessing includes: converting the original image into a grayscale image; converting the grayscale image into a binary image based on a set threshold; filling the contour of the binary image by filling holes; The image is subjected to pixel analysis to obtain the parameter vector.

Preferably, the recognition device adopts a smart phone or a tablet computer, and the recognition device obtains the classified recognition result information based on cloud computing.

The invention provides a blood detection and identification method, comprising the following steps:

Step 1. Pass the blood and hydrogel precursor solution into the chip chamber of the microfluidic chip through the chip inlet of the microfluidic chip, and the microhydrogel column is arranged in the chip chamber, and the Blood cells solidify in the micro-hydrogel column to make a hydrogel actuator;

Step 2, acquiring the original image of blood cells through the image acquisition and transmission component acquisition;

Step 3, pressing the mechanical transmission component, and transmitting the force generated by external pressing to the hydrogel actuator through the mechanical transmission component, so that the blood cells in the hydrogel actuator are braked;

Step 4, collecting and acquiring blood cell deformation images through the image acquisition and transmission component;

Step 5. Transmitting the original blood cell image and the deformed blood cell image to a recognition device through the image acquisition and transmission component;

Step 6. Obtain classified recognition result information through the recognition device.

Preferably, said step 6 includes the following sub-steps:

Step 6.1, preprocessing the obtained original image of blood cells and deformed image of blood cells by the image preprocessing module in the recognition device to obtain image vectors and parameter vectors of blood cells; the image vectors include original image vectors and deformed image vectors, The parameter vector includes a morphological parameter vector and a mechanical parameter vector;

Step 6.2, train and optimize the pre-built blood detection and classification model through the training optimization module in the identification device, and obtain the trained blood detection and classification model;

Step 6.3: After preprocessing the original image of the blood cell to be classified and recognized and the corresponding deformed image of the blood cell through the classification recognition module in the recognition device, input it into the trained blood detection and classification model to obtain the classification and recognition result information;

Wherein, the blood detection classification model is constructed based on a deep convolutional neural network, and the blood detection classification model includes six convolutional layers and three fully connected layers; the three fully connected layers contain 40, 64 and 20 vectors respectively , where the first fully connected layer contains 32 image vectors and 8 parameter vectors;

The recognition device obtains the classified recognition result information based on cloud computing.

One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

In the invention, the provided blood tester utilizes the hydrogel actuator to accurately control the blood cells in the blood to deform and maintain stable detection through continuous pressure, which can realize high-precision, easy-to-operate, low-cost blood testing and blood testing. quality monitoring. Combined with the identification equipment, the classification and identification result information can be obtained according to the original image of blood cells and the deformed image of blood cells, so as to achieve the effect of accurate identification of blood cells.

Description of drawings

FIG. 1 is a schematic structural diagram of a blood detection and identification system provided by an embodiment of the present invention;

Fig. 2 is a comparison chart of monitoring performance between a blood detection and identification system provided by an embodiment of the present invention and a laser diffraction red blood cell deformation analyzer in the prior art;

Fig. 3 is a schematic diagram of a blood detection and recognition method provided by an embodiment of the present invention combining neural network and cloud computing.

Among them, 1-microfluidic chip, 2-mechanical transmission component, 3-chip light source module, 4-optical lens, 5-image transmission module, 6-image acquisition module, 7-upper cover, 8-lower cover Department, 9-fixture department;

11-chip inlet, 12-chip outlet, 13-chip chamber, 14-micro hydrogel column;

21-spring, 22-button, 23-pad.

detailed description

In order to better understand the above-mentioned technical solution, the above-mentioned technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

Example 1:

Embodiment 1 provides a blood tester, as shown in FIG. 1 , including: a microfluidic chip 1 , a mechanical transmission component 2 , an image acquisition transmission component and a casing. The microfluidic chip 1 includes a chip inlet part 11 , a chip outlet part 12 , a chip chamber 13 and a micro hydrogel column 14 . The microfluidic chip 1 and the image acquisition transmission assembly are installed inside the housing; the mechanical transmission assembly 2 is installed on the housing and arranged above the micro hydrogel column 14 . The microfluidic chip 1 is used to provide an imaging platform for blood cells; blood and hydrogel precursors are passed into the chip chamber 13 through the chip inlet part 11, and the micro hydrogel column 14 is located in the In the chip chamber 13, blood cells are solidified in the micro-hydrogel column 14 by exposure to blue light to form a hydrogel actuator. The force transmission component 2 is used to transmit the force generated by external pressing to the hydrogel actuator, so that the blood cells in the hydrogel actuator brake. The image acquisition and transmission component is used to acquire the original blood cell image and the blood cell deformation image, and transmit the blood cell original image and the blood cell deformation image to an external identification device.

Wherein, the mechanical transmission assembly 2 includes: a spring 21 , a button 22 and a washer 23 . One end of the spring 21 is connected with the gasket 23, and the spring 21 is arranged above the gasket 23; the button 22 is connected with the other end of the spring 21, and the button 22 is arranged on the gasket 23. Above the spring 21; the lower surface of the gasket 23 is in close contact with the upper surface of the micro hydrogel column 14.

The image acquisition and transmission assembly includes: a patch-type light source module 3 , an optical lens 4 , an image acquisition module 6 and an image transmission module 5 . The patch-type light source module 3 is arranged above the microfluidic chip 1, and the optical lens 4, the image acquisition module 6 and the image transmission module 5 are all arranged on the microfluidic chip 1. Below, the image acquisition module 6 is connected to the image transmission module 5 and the optical lens 4 respectively. The patch-type light source module 3 is used to provide a light source for blood cell imaging; the optical lens 4 is used to adjust the focus of blood cell imaging; the image acquisition module 6 is used to acquire the original image of the blood cell and the deformation of the blood cell Image; the image transmission module 5 is used to transmit the original blood cell image and the deformed blood cell image to an external identification device.

The housing includes an upper cover part 7 , a lower cover part 8 and a clamp part 9 . The upper cover part 7 and the lower cover part 8 constitute the main space of the housing, the microfluidic chip 1 and the image acquisition and transmission assembly are arranged in this space, and the clamp part 9 is installed on the upper Above the cover part 7 , the force transmission assembly 2 is installed on the clamp part 9 .

For example, the shell is made of ABS material by 3D printing process, the template of the microfluidic chip 1 is made by ultraviolet lithography technology, and the microfluidic chip 1 is made of polydimethylsiloxane ( PDMS, the refractive index is 1.406).

The button 22 is an acrylonitrile-butadiene-styrene copolymer with a diameter of 1.3cm; the stiffness coefficient of the spring 21 is 5N/cm; the diameter of the gasket 23 is 1cm, and a glass gasket is specifically used .

Example 2:

Embodiment 2 provides a blood detection and identification system, see FIG. 1 , including an identification device, and the blood detection instrument as described in Embodiment 1. The recognition device is connected to the blood detector through a data line; the recognition device is used to receive the original blood cell image and the blood cell deformation image, and obtain classification and recognition result information according to the blood cell original image and the blood cell deformation image.

Wherein, the recognition device includes an image preprocessing module, a training optimization module, and a classification and recognition module. The image preprocessing module is used to preprocess the obtained blood cell original image and blood cell deformation image to obtain an image vector and a parameter vector of blood cells; the image vector includes an original image vector and a deformed image vector, and the parameter vector includes a shape A vector of scientific parameters and a vector of mechanical parameters. The training optimization module is used for training and optimizing the pre-built blood detection classification model to obtain a trained blood detection classification model. The classification and identification module is used to preprocess the original image of the blood cell to be classified and identified and the corresponding deformed image of the blood cell, and then input it into the trained blood detection and classification model to obtain the classification and identification result information.

Specifically, the blood detection classification model is constructed based on a deep convolutional neural network, and the blood detection classification model includes six convolutional layers and three fully connected layers; the three fully connected layers contain 40, 64 and 20 vectors respectively , where the first fully connected layer contains 32 image vectors and 8 parameter vectors.

The preprocessing includes: converting the original image into a grayscale image; converting the grayscale image into a binary image based on a set threshold; filling the contour of the binary image with a hole filling operation; pixelating the filled image Analyze to get the parameter vector.

The recognition device adopts a smart phone or a tablet computer, and the recognition device obtains the classified recognition result information based on cloud computing.

Example 3:

Embodiment 3 provides a blood detection identification method, comprising the following steps:

Wherein, said step 6 includes the following sub-steps:

Step 6.3: Preprocess the original blood cell image to be classified and recognized and the corresponding deformed blood cell image by the classification recognition module in the recognition device, and then input it into the trained blood detection and classification model to obtain classification and recognition result information.

Wherein, the blood detection classification model is constructed based on a deep convolutional neural network, and the blood detection classification model includes six convolutional layers and three fully connected layers; the three fully connected layers contain 40, 64 and 20 vectors respectively , wherein the first fully connected layer contains 32 image vectors and 8 parameter vectors; the recognition device obtains the classification recognition result information based on cloud computing.

For example, the ratio of blood and hydrogel precursor is 1:100; exposure by blue light (Flashlight, FENIX TK25RB) under a photomask (Filin film, Jixiangguangdian).

The present invention will be further described below.

A blood detection and identification system provided by the present invention can realize accurate cell identification based on double labeling of hydrogel actuator morphology and mechanics, and utilize mechanical transmission components to manipulate hydrogel actuators, resulting in blood cell morphology ( Diameter, roundness, axial ratio and corresponding distribution width) and mechanical parameters (deformability and distribution width) are changed, and an adjustable imaging platform is designed to capture blood cell images on different focal surfaces, and the collected images are processed by recognition equipment The image of blood cells can be used to accurately identify blood cells by combining morphological and mechanical double recognition.

For example, a smartphone is used to collect blood cell images in a field of view of a microhydrogel column of 360 microns × 360 microns through an optical lens, and the image is converted into an 8-bit grayscale image; after the software automatically adjusts the light intensity and contrast, through The threshold operation converts the blood cell image into a binary image; the threshold setting is used to remove cell debris and stacking; then the hole filling operation is used to fill the contour, which improves the calculation accuracy; finally, the area and perimeter are calculated by software pixel analysis. Precise cell identification is assisted by multivariate morphological parameters (diameter, roundness, axial ratio and corresponding distribution width) and mechanical parameters (deformability and distribution width).

The present invention realizes blood classification and recognition based on deep learning by introducing cloud computing, takes the data (including morphological and mechanical double variable parameters) integrated by smart phones as the input of cloud computing, and then converts them into vector tables, and loads Into the image vector, the classification and identification of blood disease types are performed according to the trained neural network. Among them, deep convolutional neural networks can handle a flexible number of input images.

Cloud Computing Based on Deep Learning: First, the input image is obtained after data augmentation of the original image by random rotation, cutting and flipping. Then use AlexNet with pre-trained weight coefficients on the ImageNet dataset for training. The layer before the last layer is set to have 32 neurons to meet the accuracy and time-consuming requirements. After training, these vectors are extracted as feature vectors in the embedding space learned by the neural network. All inputs are resized to 224 x 224 pixels for imaging cells in hydrogel actuators, the scale of the images is 0.1 mm x 0.1 mm, and the images should be cropped to the same scale before being used as input - 0.1mm x 0.1mm. The 32-dimensional feature vectors extracted in the images were combined with the 8-dimensional mechanical and morphological data obtained from the imaging analysis and the pooled feature vectors to train a neural network with two hidden fully-connected layers. Dropout is used during training to avoid overfitting and improve generalization performance.

In order to verify the monitoring function of the present invention, the blood detection and identification system provided by the present application is compared with the red blood cell deformation analyzer based on laser diffraction in the prior art in terms of monitoring performance of blood cell deformation, see FIG. 2 . Specifically, Fig. 2a shows the RBC elongation index value (EI, average=0.5373, σ=0.0082) obtained by laser diffraction red blood cell deformation analyzer (Lorrca, shear stress=6Pa) for 63 healthy red blood cell samples and a scatter plot of hydrogel actuator-based erythrocyte deformability values (Dr, mean = 1.210, σ = 0.0046) obtained by the present invention (stress = 3 kPa). Figure 2b shows that by Bland-Altman analysis, the average deviation of the erythrocyte deformability values based on the hydrogel actuator and the laser diffraction erythrocyte deformation analyzer is 0.6723, and the SD is 0.0059. The limit of anastomosis (LOA) was between 0.6607-0.6840. Figure 2c shows the mountain diagram analysis calculated percentiles for each ranking difference in cell deformability between the laser diffraction method and the hydrogel actuator method, with the vertical dashed line representing the center of the mountain and the horizontal dashed line representing the first 5 to 95th percentile. Figure 2 illustrates that the device provided by the present invention maintains high precision while being miniaturized.

Figure 3 shows that the smartphone collects the image and multivariate (morphology: diameter, roundness, axial ratio and corresponding distribution width; mechanics: deformability and distribution width) integrated and loaded to the cloud, based on the trained neural network. For classification recognition, see Figure 3a to Figure 3d, the classification recognition data is returned to the smartphone; patients can share classification recognition results and detailed parameters with doctors through cloud sharing. Referring to Figure 3e to Figure 3j, five typical blood diseases and corresponding models under healthy conditions were constructed. The five typical blood diseases include megaloblastic anemia (MA), myelofibrosis (MF), iron deficiency anemia (IDA), thrombocytopenic purpura (TTP) and thalassemia (THAL.). Figure 4j shows the performance of the blood detection classification model after training with three training methods, the first is morphological training, the second is morphological combined with image training, and the third is combined morphological, mechanical and image training . By comparison, it can be seen that the classification model combined with image, morphological and mechanical parameters achieved 100% classification recognition accuracy in six classifications (n=432).

In summary, the present invention can realize high-precision, easy-to-operate, low-cost blood detection and blood quality monitoring. Combined with the identification equipment, the classification and identification result information can be obtained according to the original image of blood cells and the deformed image of blood cells, so as to achieve the effect of accurate identification of blood cells.

Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention without limitation, although the present invention has been described in detail with reference to examples, those of ordinary skill in the art should understand that the technical solutions of the present invention can be carried out Modifications or equivalent replacements without departing from the spirit and scope of the technical solutions of the present invention shall be covered by the claims of the present invention.

Claims

A blood detector, characterized in that it includes: a microfluidic chip, a mechanical transmission component, an image acquisition and transmission component, and a housing;

The microfluidic chip includes a chip inlet, a chip outlet, a chip chamber and a micro hydrogel column;

The microfluidic chip and the image acquisition transmission assembly are installed inside the housing; the mechanical transmission assembly is installed on the housing and arranged above the micro hydrogel column;

The microfluidic chip is used to provide an imaging platform for blood cells; blood and hydrogel precursors are passed into the chip chamber through the chip inlet, and the micro hydrogel column is located in the chip chamber , curing blood cells in the micro-hydrogel column by exposure to blue light to make a hydrogel actuator;

The force transmission component is used to transmit the force generated by external pressing to the hydrogel actuator, so that the blood cells in the hydrogel actuator brake;

The image acquisition and transmission component is used to acquire the original blood cell image and the blood cell deformation image, and transmit the blood cell original image and the blood cell deformation image to an external identification device.
The blood testing instrument according to claim 1, wherein the mechanical transmission component comprises: a spring, a button and a washer;

One end of the spring is connected to the gasket, and the spring is arranged above the gasket; the button is connected to the other end of the spring, and the button is arranged above the spring; the pad The lower surface of the sheet is in close contact with the upper surface of the micro hydrogel column.
The blood detector according to claim 1, wherein the image acquisition and transmission component comprises: a patch-type light source module, an optical lens, an image acquisition module, and an image transmission module;

The patch light source module is arranged above the microfluidic chip, the optical lens, the image acquisition module and the image transmission module are all arranged under the microfluidic chip, and the image acquisition The modules are respectively connected with the image transmission module and the optical lens;

The patch-type light source module is used to provide a light source for blood cell imaging; the optical lens is used to adjust the focus of blood cell imaging; the image acquisition module is used to acquire the original image of blood cells and the deformed image of blood cells; The image transmission module is used to transmit the original blood cell image and the deformed blood cell image to an external identification device.
A blood detection and identification system, characterized by comprising: an identification device, and the blood testing instrument according to any one of claims 1-3;

The identification device is connected with the blood detector through a data line; the identification device is used to receive the original image of blood cells and the deformed image of blood cells, and obtain the classification and identification result information according to the original image of blood cells and the deformed image of blood cells.
The blood detection and recognition system according to claim 4, wherein the recognition device includes an image preprocessing module, a training optimization module, and a classification and recognition module;

The image preprocessing module is used to preprocess the obtained blood cell original image and blood cell deformation image to obtain an image vector and a parameter vector of blood cells; the image vector includes an original image vector and a deformed image vector, and the parameter vector includes a shape The vector of scientific parameters and the vector of mechanical parameters;

The training optimization module is used to train and optimize the pre-built blood detection classification model to obtain a trained blood detection classification model;

The classification and identification module is used to preprocess the original image of the blood cell to be classified and identified and the corresponding deformed image of the blood cell, and then input it into the trained blood detection and classification model to obtain the classification and identification result information.
The blood detection and recognition system according to claim 5, wherein the blood detection classification model is constructed based on a deep convolutional neural network, and the blood detection classification model includes six convolutional layers and three fully connected layers; The first fully connected layer contains 40, 64 and 20 vectors respectively, and the first fully connected layer contains 32 image vectors and 8 parameter vectors.
The blood detection and identification system according to claim 5, wherein the preprocessing includes: converting the original image into a grayscale image; converting the grayscale image into a binary image based on a set threshold; using the filling hole operation Filling the contour of the binary image; performing pixel analysis on the filled image to obtain the parameter vector.
The blood detection and recognition system according to claim 4, wherein the recognition device adopts a smart phone or a tablet computer, and the recognition device obtains the classified recognition result information based on cloud computing.
A blood detection and identification method, characterized in that it comprises the following steps:

Step 1. Pass the blood and hydrogel precursor solution into the chip chamber of the microfluidic chip through the chip inlet of the microfluidic chip, and the microhydrogel column is arranged in the chip chamber, and the Blood cells solidify in the micro-hydrogel column to make a hydrogel actuator;

Step 2, acquiring the original image of blood cells through the image acquisition and transmission component acquisition;

Step 3, pressing the mechanical transmission component, and transmitting the force generated by external pressing to the hydrogel actuator through the mechanical transmission component, so that the blood cells in the hydrogel actuator are braked;

Step 4, collecting and acquiring blood cell deformation images through the image acquisition and transmission component;

Step 5. Transmitting the original blood cell image and the deformed blood cell image to a recognition device through the image acquisition and transmission component;

Step 6. Obtain classified recognition result information through the recognition device.
The blood detection and identification method according to claim 9, wherein said step 6 comprises the following sub-steps:

Step 6.1, preprocessing the obtained original image of blood cells and deformed image of blood cells by the image preprocessing module in the recognition device to obtain image vectors and parameter vectors of blood cells; the image vectors include original image vectors and deformed image vectors, The parameter vector includes a morphological parameter vector and a mechanical parameter vector;

Step 6.2, train and optimize the pre-built blood detection and classification model through the training optimization module in the identification device, and obtain the trained blood detection and classification model;

Step 6.3: After preprocessing the original image of the blood cell to be classified and recognized and the corresponding deformed image of the blood cell through the classification recognition module in the recognition device, input it into the trained blood detection and classification model to obtain the classification and recognition result information;

Wherein, the blood detection classification model is constructed based on a deep convolutional neural network, and the blood detection classification model includes six convolutional layers and three fully connected layers; the three fully connected layers contain 40, 64 and 20 vectors respectively , where the first fully connected layer contains 32 image vectors and 8 parameter vectors;

The recognition device obtains the classified recognition result information based on cloud computing.