Description:Complete Specification:

The following specification describes and ascertains the nature of this invention and the manner in which it is to be performed.
Field of the invention
[0001] This invention is related to a control unit for determining anisocytosis in a blood sample and a method thereof.

Background of the invention
[0002] In the medical procedures, analysis of blood samples provides clinically significant information. Currently blood samples are analyzed using cell counters followed by peripheral smear examination. Despite the vast technological advancement, peripheral blood smear examination plays a significant role because of the detailed clinical insights on cellular anomalies, which were not analyzed in-detail, by automated flow cytometers. Peripheral Smear examination is usually performed to study the morphological abnormalities of blood components and their distribution. On average, nearly 6 billion blood tests are performed every year requiring huge potential pathologists.

[0003] A US patent application 7838296 discloses a diagnostic device is able to perform measurements on blood flow using physiological salt solution, which is harmless to living objects, as a tracer. The device has a calculation process performed on the electric signal, which represents changes in concentration of hemoglobin generated by injection of the physiological salt solution. At least one of the blood flows, the absolute concentration of oxyhemoglobin or deoxyhemoglobin is displayed.
Brief description of the accompanying drawings
[0004] Figure 1 illustrates a control unit in a digital focusing device, in accordance with an embodiment of the invention; and
[0005] Figure 2 illustrates a flowchart of a method of determining of Anisocytosis of a blood sample in accordance with the present invention.

 
Detailed description of the embodiments
[0006] Figure 1 illustrates a control unit for determining and representing Anisocytosis of a blood sample, in accordance with an embodiment of the invention. The control unit 12 captures multiple images 15 of the blood sample 14 using an automated digital image acquisition unit 16 and calibrates the capture images of the blood sample 14 to a micrometer scale. The control unit 12 converts the at least one captured image 15 of to a corresponding evenly contrasted image and detects multiple cells 18 of interest in the evenly contrasted image 15 and enhance each of the detected cell 18 using at least one signal processing technique. The control unit 12 extracts at least one feature of each cell 18 and identifies an optimized feature set from extracted features for categorizing the detected cells 18. The control unit 12 identifies anisocytosis from at least one the extracted features using at least one intelligence network and determines a number of the anisocytosis in the detected cells 18 based on a predefined feature value.

[0007] Further, the construction of the control unit 12 and the components of the control unit 12 is explained as follows. The Anisocytosis is a process of representing size variations in the red blood cells 18. The Anisocytosis represents unequal size of red blood cells in the blood sample 14. The control unit 12 is chosen from a group of control units comprising a microprocessor, a microcontroller, a digital circuit, an integrated chip and the like. The control unit 12 receives at least one signal related to the blood sample 14, wherein the signal is chosen from a group of signals comprising an image, a pulse signal and the like. However, it is to be understood that the control unit 12 receives the data related to the blood sample in any other form that is known to a person skilled in the art but is not restricted to the above-mentioned signal forms.

[0008] According to one embodiment of the invention, the signal as per the present invention is an image and the automated digital image acquisition unit 16 comprises an image processing unit which uses any one of the image processing techniques known in the state of the art. The blood of a human being is smeared on the sample 14 for the analysis /diagnosis purpose. The sample 14 having the human body fluid/specimen like blood is placed in one of the analysis devices that is known in the state of the art.

[0009] According to one embodiment of the invention, the blood sample 14 is placed in a digital focusing device 10. The digital focusing device 10 is used to focus the content of the blood sample 14 and to analyze the content of the blood sample 14 for diagnosing purpose. According to one embodiment, the control unit 12 is made as an integral component of the digital focusing device 10. The device 10 captures the at least one image 13 (which is two-dimensional) of the content of the blood sample 14 and processes it further to determine and represent the red blood cells 18 of the blood sample 14.

[0010] In another embodiment, the control unit 12 is an external source connected to the device 10, wherein the control unit 12 is a cloud repository that is connected to the device through a communication means. The control unit 12 displays the final result of the Anisocytosis/cellular size variation values or the severity of the identified at least one medical condition to the user directly using the communication means.

[0011] The artefacts 15 in the content of the blood sample 14 is identified using at least one imaging processing technique or a learning model like an artificial intelligence (AI) model, deep learning technique (DL), a machine learning (ML) technique, a blurring technique or by using any one of the linear filters and non-linear filters. The basic information on the working methodology of the AI model or the DL/ML technique is not explained in detail, as it is known in the state of the art.

[0012]The at least one extracted feature is chosen from a group of characteristics comprising a surface area, a feret, a total area of the signal, Mean, Centroid, Perimeter, Angle, Circularity, Integrated Density, Skewness, Kurtosis, Aspect ratio, Roundness, Solidity, Rectangularity, Correlation, Entropy, Elongation, Solidarity, Convexity, Axis Percentage Difference, Variance, Homogeneity, Prominence, Shade, Energy, a cellular geometry , a morphology, color, texture, a cellular arrangement and the like. However, the extracted features are not restricted to the above disclosed features but can be any other features that provides the information of at least one cell of the human body fluid (which is a blood sample according to one embodiment of the present invention).

[0013] Figure 2 illustrates a flowchart of a method of determining and representing Anisocytosis from a signal 13 related to a blood sample 14 in accordance with the present invention. In step S1, multiple images 15 of the blood sample 14 are captured using an automated digital image acquisition unit 16 of a control unit12. In step S2, the captured images of the blood sample 14 are calibrated to a micrometer scale by the control unit 12. In step S3, at least one captured image 15 is converted into a corresponding evenly contrasted image. In step S4, multiple cells 18 of interest are detected in the evenly contrasted image 15 and each of the detected cell 18 is enhanced using at least one signal processing technique. In step S5, at least one feature of each cell 18 is extracted and an optimized feature set is identified from extracted features for categorizing the detected cells 18. In step S6, anisocytosis from at least one the extracted features are identified using at least one intelligence network 20 and a number of the anisocytosis in the detected cells 18 are determined based on a predefined feature value. Based on the above disclosed information, the control unit 12 detects a severity of at least one medical condition.

[0014] The above method is explained in detail. For better understanding of the invention, the device 10 is considered as the digital focusing device 10, the content of the sample 14 is a blood sample and the control unit captures multiple images of the blood sample 14 for further analysis. In the present invention, the control unit 12 is determining anisocytosis in the given blood sample 14 of the patient using the above digital focusing device 10. The method also explains identification of the different types of the red blood cells 18 in the blood sample 14 based on the cellular size of each red blood cell 18 in the given blood sample 14. It is to be noted that, the cellular size is obtained from the at least one feature value of the optimized feature set.

[0015] The blood on the sample 14 is smeared and is placed in the digital focusing device 10. The control unit 12 captures multiple images 15 of the blood sample 14 using the automated digital image acquisition unit 16 (e.g.: a camera) of the device 10. At least one image 15 comprises different cells 18 of the blood and the artefacts. The control unit 12 removes artefacts from the image 15 of the blood sample 14 using the signal-processing unit. The control unit 12 detects multiple cells of interest in the at least one image 15 and enhances each of the detected cell 18 using at least one signal processing technique. The control unit 12 labels each of the detected cells 18 after enhancing each of the detected cell 18.

[0016] Using at least one image processing technique, the control unit 12 removes all the artefacts present in the image 15. In other words, the control unit 12 differentiates between the cells 18 of interest and the artefacts using any one of the following processing techniques like AI model, Machine learning technique, Deep learning technique and the like. The size of the blood cells is measured in micrometer scale. In order to match the standard measurement scale, the captured images 15 are calibrated to a micrometer scale by the control unit 12. The Image processing unit 16 modifies any magnification value with a configurable parameter. The control unit 12 upon detecting the multiple cells 18 of interest enhances the cells 18 by enhancing the edges of the cells 18.

[0017] The cells of interest 18 with the enhanced edges are used for further processing in determining the cellular size of the red blood cells in the blood sample. According to one embodiment of the invention, instead of addressing a specific problem like artefacts removal, uneven illumination, uneven contrast, out of focus or blurredness in at least one captured image, the image processing unit of the control unit preserves and enhance the edges of the detected cells while suppressing the noises in the background. The image processing unit converts an input image of any varied image quality to an evenly contrasted image where there is clear distinction between the cells in the foreground and a background of the image 15.

[0018] The control unit 12 isolates the enhanced cells of interest from the background of the image 13 and labels each cell 18. For instance, the control unit 12 labels each cell 18 as L1, L2, and L3…Ln. The control unit 12 upon labelling each cell 18 extracts at least one feature (as mentioned above) of each cell 18. The control unit 12 extracts the features from each of the labelled cells 18.

[0019] The control unit is fed with an artificial intelligence model (AI), a machine-learning module (ML), a deep learning model (DL), which is built based on the extracted feature sets. The real-time features are extracted and then fed to the model for further analysis. The explanation on the learning module like artificial intelligence model (AI), machine learning module (ML), deep learning models (DL) and the like is not provided much in detail, as it is known in the state of the art. The features as mentioned above are used to differentiate the cells 18 into different categories. According to one embodiment of the invention, the control unit 12 categorizes the labelled cells 18 into at least two categories comprising red blood cells (RBC) 18 and the non-red blood cells (non-RBC). The Non-RBC comprises white blood cells, the blood platelets and the others.

[0020] The control unit 12 upon categorizing the labelled cells 18 based on the extracted features, determines the count of the categorized red blood cells 18 and the count of the non-red blood cells. The count of red blood cells 18 is considered as “n” and the count of non- red blood cells is considered as “n1”.

[0021] The control unit 12 extracts multiple features from each of the identified labelled cell. For instance, the control unit 12 extracts more than 164 features related to each of the labelled cell from the captured image. It is to be noted that the labelled cell and the detected cell holds the same meaning. The control unit 12 identifies an optimized feature set from the extracted features. The classification accuracy depends on the efficiency of the features in differentiating the anomaly types. The selection of appropriate feature highly depends on the unique characterization of each anomaly from the others. The optimized feature set is selected based on the morphological, geometrical features, Morphology, Cellular Arrangement, Color, Texture and Inclusions was employed to build a Cascaded intelligence (e.g.: AI) Architecture. Based on the targeted cell anomalies, the control unit 10 divides the cells into macrocyte, macrocyte and normocyte.

[0022]The optimized feature set comprises the features like a cellular size/area, a pallor size/area, a kurtosis value, an energy value, difference pallor, rectangular difference pallor, axis percentage difference pallor, feret major difference pallor, feret minor difference pallor, Chroma decider pallor, circular area difference pallor, red center mean, blue center mean, green center mean, target blob, compactness pallor, form factor pallor, area, area pallor, red mean pallor, green mean pallor, blue mean pallor, compactness pallor, rectangularity pallor, elongation pallor, circularity pallor, aspect ratio pallor and the like. However, it is to be noted that, the optimized feature set is not restricted to the above features, but can be of any other features as known to a person skilled in the art.

[0023] The control unit 12 identifies different types of red blood cells 18 and cellular size of the red blood cells 18 from at least one of the above-disclosed extracted features of the optimized set. The control unit 12 determines the number of each of the identified type of red blood cell 18 based on a predefined feature value. The predefined feature value is fed to the device 10 using at least one learning techniques comprising an artificial intelligence (AL) model, a machine-learning module (ML), a deep learning model (DL) or the like. For instance, at least one feature is a cellular size, and the predefined feature value is the cellular size of an ideal red blood cell.

[0024] It is to be noted that, at least one feature is not restricted to only cellular size but can be of any other feature comprising Feret, Surface area, area and the like. The control unit 12 identifies anisocytosis from at least one of the extracted features using at least one intelligence network and determine a number of the anisocytosis in the detected cells 18 based on a predefined feature value. In the extracted features, highly significant latent features are selected while forming the optimized feature set. The optimized feature set comprises features extracted specific to characterization of anisocytosis which includes Area, Major Axis Length, Minor Axis length, Roundness, Solidity, Aspect Ratio, Bounding Rect width, Bounding Rect Height, Rectangularity, Feret, Minimum Feret and the like. The control unit 12 detects the number of anitocystosis cells based on the total number of macrocytes, microcytes and normocytes in the detected cells 18.

[0025] The predefined feature value is stored in the control unit 12 of the device 10 during the calibration process. The control unit 12 compares the cellular size of the each of the red blood cell 18 present in the blood sample 14 with the predefined cellular size value. Based on the comparison result the control unit 12 identifies the types of red blood cells 18 as microcytic red blood cell, normocytic red blood cell, macrocytic red blood cell and the like. If the predefined cellular size value is X and the calculated cellular size (X1) of one red blood cell 18 is more than the predefined cellular size value (i.e., X1>X, wherein X1 being the calculated cellular size and X being the predefined cellular size), then the control unit 12 identifies the respective cell as Macrocytic cell. If the calculated cellular size of one red blood cell 18 (X2) is less than the predefined cellular size (i.e., X2< X), then the control unit 12 identifies the cell as Microcytic cell. If the calculated cellular size of one red blood cell (X3) is same as the predefined cellular size (i.e., X3=X), then the control unit 12 identifies the cell as Normocytic cell. The type is not restricted to the above-mentioned red blood cells 18 but can any other type that is known in the state of the art.

[0026] The above-determined at least one feature value (cellular size), is determined from the image of the blood sample 14. The control unit 12 determines at least one feature values from the image 15 (which is a two-dimensional signal) of the blood sample 14, which is a simple and cost –effective solution when compared to the conventional methods. It also represents the identified type of red blood cells based on the determined at least one feature values and the severity of at least one medical condition. With the above method and control unit 12 in the device 10, many blood characteristics are effectively determined. The present invention is highly compliant to handle myriad of raw image data variations including image intensity fluctuations, different staining conditions, myriad of smearing quality.

[0027] Even in the cases of custom-made smear of the content of the sample 14 or in the conventional methods, the present invention provides a robust solution in determining at least one characteristic of a human body fluid in detecting at least one abnormality in the human body. The above-disclosed method evades the necessity of high-rate 360-degree rotational movement of the content of the sample 14, at approximately 3200 –3500 rpm for 15 –20 minutes. This centrifugation of the blood sample 14 will have an impact on the clinical reports being generated. The present invention does not require a controlled environment to be established.

[0028] The present invention helps in avoiding the reagent cost used in cell counters, which are currently used for volumetric analysis of blood. With the above disclosed method, estimation of both volumetric indices/ count metrics and cellular morphological findings from single system is achieved, thereby overcoming the necessity of two independent devices (as in the conventional systems) and reducing the cost & system maintenance. The method disclosed in the present invention eliminates the lysing of cells, thereby overcoming the necessity of chemicals in the volumetric analysis and avoids the chemicals used for retaining the cellular morphology.

[0029] With the present invention disclosing a method of determining the at least one feature value and cellular variation values of the blood sample 14 and representing the same in the device 10 in the form of a graphical representation, at least one medical condition can be easily identified. The medical condition can be detected from a two-dimensional image signal through at least one image processing technique and through at least one learning module (e.g.: artificial Intelligence (AI) model). A mathematical machine-learning hypothesis is being built combining both the statistical hypothesis and clinical hypothesis in order to detect the cellular variation from the two-dimensional vector.

[0030] The above-disclosed method provides a solution in reducing the cost function by adapting several function approximation techniques and by assigning high weights to the significant feature vectors thus eliminating the feature vectors resulting in null hypothesis. The above-disclosed optimized methodology manifests the two-dimensional digital space eligible for employing the clinically accepted red cell indices formula. In addition to this, the proposed methodology employs techniques to calculate at least one feature value and representing the severity of the medical conditions based on the determined feature value and cellular variation values in the form of a graphical representation.

[0031] It should be understood that embodiments explained in the description above are only illustrative and do not limit the scope of this invention. Many such embodiments and other modifications and changes in the embodiment explained in the description are envisaged. The scope of the invention is only limited by the scope of the claims
, Claims:We Claim:

1. A control unit (12) for determining anisocytosis of red blood cells (18) present in a blood sample (14), said control unit (12) adapted to:
- capture multiple images (15) of said blood sample (14) using an automated digital image acquisition unit (16);

characterized in that:
- calibrate said capture images (15) of said blood sample (14) to a micrometer scale;
- convert said at least one captured image (15) of to a corresponding evenly contrasted image;
- detect multiple cells (18) of interest in said evenly contrasted image (15) and enhance each of said detected cell (18) using at least one signal processing technique;
- extract at least one feature of said each cell (18) and identify an optimized feature set from extracted features for categorizing said detected cells (18);
- identify anisocytosis from at least one said extracted features using at least one intelligence network (20) and determine a number of said anisocytosis in said detected cells (18) based on a predefined feature value.

2. The control unit (12) as claimed in claim 1, wherein said control unit (12) adapted to calibrate said at least one captured image (15) to said micrometer scale for detecting a size of each of said detected cells (18).

3. The control unit (12) as claimed in claim 1, wherein said detected cells (18) are categorized into multiple groups comprising a microcytic group, a macrocytic group and a normocytic group.

4. The control unit (12) as claimed in claim 1, wherein an optimized feature set is chosen from said extracted features, said optimized featured set comprises features like a Area, Major Axis Length, Minor Axis length, Roundness, Solidity, Aspect Ratio, Bounding Rect width, Bounding Rect Height, Rectangularity, Feret, Minimum Feret.

5. The control unit (12) as claimed in claim 4, wherein determines a volume of each of said at least one categorized cell from said at least two extracted features of said optimized set, for determining at least one medical condition and at least one volume characteristic of said blood sample (14).

6. The control unit (12) as claimed in claim 1, wherein said control unit (12) adapted to distinguish cells (18) present in a foreground of said at least one captured image (15) and a background of said at least one captured image (15) using said at least one processing technique.

7. The control unit (12) as claimed in claim 1, wherein said intelligence networks (20) are chosen from a group of intelligence networks comprising an artificial intelligence (AI) network, a machine learning network (ML), a deep learning network (DL).

8. A method of determining anisocytosis of red blood cells (18) present in a blood sample (14), said method comprising:
- capturing multiple images (15) of said blood sample (14) using an automated digital image acquisition unit (16) of a control unit (12);

characterized in that:
- calibrating said capture images (15) of said blood sample (14) to a micrometer scale by said control unit (12);
- converting said at least one captured image (15) of to a corresponding evenly contrasted image;
- detecting multiple cells (18) of interest in said evenly contrasted image (15) and enhancing each of said detected cell (18) using at least one signal processing technique;
- extracting at least one feature of said each cell (18) and identifying an optimized feature set from extracted features for categorizing said detected cells (18);
- identifying anisocytosis from at least one said extracted features using at least one intelligence network (20) and determining a number of said anisocytosis in said detected cells (18) based on a predefined feature value.