N-BGP (Noninvasive Blood Group Prediction Dataset)

Citation Author(s):
Nitin
Ujgare
Aryan
Verma
Prem Kumari
Verma
Nagendra Pratap
Singh
Submitted by:
Prem Verma
Last updated:
Wed, 07/05/2023 - 06:55
DOI:
10.21227/81ps-bx03
Data Format:
License:
5
1 rating - Please login to submit your rating.

Abstract 

This Dataset used a non-invasive blood group prediction approach using deep learning. Rapid and meticulous prediction of blood type is a major step during medical emergency before supervising the red blood cell, platelet, and plasma transfusion. Any small mistake during transfer of blood can cause death. In conventional pathological assessment, the blood test is conducted using automated blood analyser; however, it results into time taking process. In typical pathological blood test, when blood sample is collected by pricking the skin, it may originate bleeding, lead to fainting and can cause skin laceration at certain part of body. The proposed deep learning approach is a non-invasive without perforating the skin that automatically predict the human blood type by applying deep learning algorithms to the captured images of superficial blood vessels present on finger. As laser light passes, the optical image of blood vessels hidden on the finger skin surface area is captured, which incorporates specific antigen shapes such as antigen ‘A’ and antigen ‘B’ present on the surface. Captured shapes of different antigen further used to predict the blood group of humans. The system requires high-definition camera to capture the antigen pattern from the red blood cells surface for classification of blood type without piercing the skin of patient. The proposed solution is easy to implement, straightforward and significant to identify ABO blood group instantly. It provides cost effective solution for identifying blood group in case of medical emergencies, battleground and more useful for infants.

  

Instructions: 

The dataset contains images of blood group samples of 103 participants. These Blood group samples are A+, A-, AB+, O+ and O-. These participants submitted the samples through the device which is made by author for collection of blood sample. The main file of the dataset is named A+.zip, A-. Zip, AB+.Zip, O-. Zip and AB+.Zip .This file contains male and female both blood samples. The total number of images is 103. All the images in .png format.

A desktop application is implemented in python that reads the users demographic details and captures the images through above black box containing laser light and web camera.

The data collection process is carried out as follows

1. The user will place his/her thumb under the light source.

2. The user’s finger will directly come under the proximity of light source which will be a laser

light.

3. Below the hand, the device capturing the images which is typically a web camera or mobile

camera is placed.

4. When laser light is turned on, the light is illuminated at certain frequencies and absorbed by

the haemoglobin in the red cells, perhaps due to coherent nature of illumination, the light gets

scattered in a small margin on hitting the edges of the antigenic determinants having a

specific structure. The scattering of laser light by antigens or antigenic determinants provides

valuable information about their spatial distribution and properties.

5. Camera device captures multiple images for different light scatterness pattern for a certain

 

period to obtain after-effect of scattering.