Smartphone Based Automatic Diagnostic Healthcare System

Vaish, Pallavi and P, Rajalakshmi (2018) Smartphone Based Automatic Diagnostic Healthcare System. Masters thesis, Indian Institute of Technology Hyderabad.

Text
Thesis_Mtech_EE_3735.pdf - Submitted Version
Restricted to Repository staff only until January 2020.
Download (7MB) | Request a copy

Abstract

This thesis mainly focuses on the smartphone based implementation of Organ validation and automated Computer Aided Diagnosis (CAD) of ultrasound images of kidney and liver. The transmission of the acquired images and videos from the ultrasound scanner from remote areas to the doctors sitting in urban areas for getting diagnosis is termed as Telesonography. Telesonography suffers from inherent limitations due to the need of all time availability of experts in cloud and data connectivity to the device. Due to the lack of trained sonographers in remote areas, the ultrasound videos scanned by untrained persons present in rural area, consist of lot of redundant information which should be removed before sending it for further processing. For the further processing of abnormality detection to be correct, organ validation is needed. CAD used for automatic detection of abnormalities without manual intervention can overcome these limitations. Commercially available ultrasound scanners restrict the installation of new softwares and hence CAD algorithms cannot be integrated into the existing ultrasound scanners. There is a need for an external computing device, which can acquire image data from ultrasound scanners, perform CAD and generate result. Smartphones are now widely used in personalized healthcare due to its ubiquitous computing capability. Smartphones with embedded CAD can be used as a computing device for automated diagnosis. We have developed Android based applications which can be used for organ validation of recorded ultrasound video and can perform CAD on the acquired ultrasound image to detect the abnormality present in the ultrasound image. The proposed algorithm for Organ Validation in acquired ultrasound videos includes extraction of GIST features from the fixed pre-defined Region Of Interest (ROI) followed by Support Vector Machine (SVM) classifier with Quadratic kernel. In case of kidney the developed application can detect whether the kidney is in normal state or abnormal state which is caused due to the presence of cyst or stone. The proposed algorithm for abnormality detection in kidney ultrasound images uses Viola Jones algorithm for the ROI detection, followed by texture feature extraction (Histogram features, GLCM features, and GLRLM features) and further it uses SVM classifier with RBF kernel to classify the images as normal or abnormal. On the other hand, from ultrasound image of liver the developed Android based application can detect the type of abnormality present in the liver which is caused due to the excess accumulation of fats in the liver cells termed as fatty liver. It can detect three kinds of fatty liver states which are Grade 1, Grade 2, and Grade 3. The proposed algorithm uses Wavelet domain features extraction followed by SVM classifier (One versus All) with RBF kernel for the detection of different kinds of abnormalities present in the liver ultrasound image. The application developed for kidney state detection resulted with an accuracy of 90.91% while the application developed for liver state detection resulted with an accuracy of 94.28% and the application developed for organ validation resulted with an accuracy of 94.93%.

[error in script]

IITH Creators: