
Praveen Gurunath Bharathi
Postdoctoral Scholar, Radiology
Bio
Praveen joined the Nuclear Medicine and Molecular Imaging, Department of Radiology as a Postdoctoral Research Fellow where he will be working on predictive and real-time PET image quality monitoring using Machine learning. Previously, he worked as a Postdoctoral Research Associate at the University of Manchester, UK where he developed an inexpensive imaging & automated analysis system to support early diagnosis of Systemic Sclerosis. He received his Ph.D. in Electrical & Electronics Engineering from the Birla Institute of Technology and Science (BITS), Pilani. His Ph.D. research was focused on developing an automated intelligent decision support system for multiple brain disorder diagnosis from MRI scans. His research is mainly in the interdisciplinary field of deep learning and medical image analysis. In particular the application of deep learning models to improve disease prognosis. He completed his Master's in Biomedical Signal Processing & Instrumentation from RV College of Engineering, India. During his master's he worked as a project intern in the Department of Aerospace Engineering at the Indian Institute of Science (IISc, Bangalore) where he worked on the application of 3D digital image correlation to perform strain analysis on the human body during various physical activities.
Professional Education
-
Master of Technology, Visveswaraiah Technology Univ (2012)
-
Bachelor of Engineering, Visveswaraiah Technology Univ (2010)
-
Doctor of Philosophy, Birla Institute of Technology and Science (2020)
-
Doctor of Philosophy, BITS Pilani - India, Electrical and Electronics Engineering (2019)
-
Master of Technology, R.V College of Engineering - India, Biomedical Signal Processing & Instrumentation (2011)
-
Bachelor of Engineering, NMAMIT - India, Computer Science Engineering (2009)
Research Interests
-
Brain and Learning Sciences
-
Data Sciences
All Publications
-
A deep learning system for quantitative assessment of microvascular abnormalities in nailfold capillary images
RHEUMATOLOGY
2023
Abstract
Nailfold capillaroscopy is key to timely diagnosis of systemic sclerosis (SSc), but is often not used in rheumatology clinics because the images are difficult to interpret. We aimed to develop and validate a fully automated image analysis system to fill this gap.We mimicked the image interpretation strategies of SSc experts, using deep learning networks to detect each capillary in the distal row of vessels and make morphological measurements. We combined measurements from multiple fingers to give a subject-level probability of SSc.We trained the system using high-resolution images from 111 subjects (Group A) and tested on images from subjects not in the training set: 132 imaged at high-resolution (Group B); 66 imaged with a low-cost digital microscope (Group C). Roughly half of each group had confirmed SSc, half were healthy controls or had primary Raynaud's phenomenon ('normal'). We also estimated the performance of SSc experts.We compared automated SSc probabilities with the known clinical status of patients (SSc versus 'normal'), generating receiver operating characteristic curves (ROCs). For Group B, the area under the ROC (AUC) was 97% [94% - 99%] (median [90% confidence interval]), with equal sensitivity/specificity 91% [86% - 95%]. For Group C, AUC was 95% [88% - 99%], with equal sensitivity/specificity 89% [82% - 95%]. SSc expert consensus achieved sensitivity 82%, specificity 73%.Fully automated analysis using deep learning can achieve diagnostic performance at least as good as SSc experts, and is sufficiently robust to work with low-cost digital microscope images.
View details for DOI 10.1093/rheumatology/kead026
View details for Web of Science ID 000921892700001
View details for PubMedID 36651676
-
Elevating Amodal Segmentation Using ASH-Net Architecture for Accurate Object Boundary Estimation
IEEE ACCESS
2023; 11: 83377-83389
View details for DOI 10.1109/ACCESS.2023.3301724
View details for Web of Science ID 001047225400001
-
Nailfold capillaroscopy: a survey of current UK practice and 'next steps' to increase uptake among rheumatologists
RHEUMATOLOGY
2022
View details for DOI 10.1093/rheumatology/keac320
View details for Web of Science ID 000818851500001
-
OA08 Development of an automated deep learning-based system for distinguishing between ‘systemic sclerosis' and ‘normal'capillaries
2022
View details for DOI 10.1093/rheumatology/keac132.007
-
P117 Nailfold capillaroscopy: a survey of current UK practice and ‘next steps’ to facilitate generalised uptake
2022
View details for DOI 10.1093/rheumatology/keac133.116
-
Combination of hand-crafted and unsupervised learned features for ischemic stroke lesion detection from Magnetic Resonance Images
BIOCYBERNETICS AND BIOMEDICAL ENGINEERING
2019; 39 (2): 410-425
View details for DOI 10.1016/j.bbe.2019.01.003
View details for Web of Science ID 000472750700011
-
Brain abnormality detection using template matching
BIO-ALGORITHMS AND MED-SYSTEMS
2018; 14 (4)
View details for DOI 10.1515/bams-2018-0029
View details for Web of Science ID 000454330900001
-
Ischemic stroke lesion segmentation using stacked sparse autoencoder
COMPUTERS IN BIOLOGY AND MEDICINE
2018; 99: 38-52
Abstract
Automatic segmentation of ischemic stroke lesion volumes from multi-spectral Magnetic Resonance Imaging (MRI) sequences plays a vital role in quantifying and locating the lesion region. Most existing methods mainly rely on designing hand-crafted features followed by a classifier model for ischemic stroke lesion segmentation. Design of these features requires complex domain knowledge and often lacks the ability to differentiate between the stroke lesions and the normal classes. In this work, we propose an unsupervised featured learning approach based on stacked sparse autoencoder (SSAE) framework for automatically learning the features for accurate segmentation of stroke lesions from brain MR images. A deep architecture is designed using sparse autoencoder (SAE) layers, followed by support vector machine (SVM) classifier for classifying the patches into normal or lesions. We validated our approach on a publicly available Ischemic Stroke Lesion Segmentation (ISLES) 2015 dataset, with a mean precision of 0.968, mean dice coefficient (DC) of 0.943, mean recall of 0.924 and mean accuracy of 0.904. The experimental results show that our proposed approach significantly outperforms the state-of-the-art methods in terms of precision, DC, and recall. Quantitative evaluation was carried out and compared with the existing approaches, which demonstrates that the proposed method is 25.71%, 36.67%, and 16.96% higher in terms of precision, DC and recall values, respectively. The unsupervised features learned via SSAE framework performs better than the hand-crafted features and can be easily trained on large datasets.
View details for DOI 10.1016/j.compbiomed.2018.05.027
View details for Web of Science ID 000442978700004
View details for PubMedID 29883752
-
MediCloud: Cloud-Based Solution to Patient’s Medical Records
Proceedings of the International Conference on ISMAC in Computational Vision and Bio-Engineering 2018
Springer, Cham. 2018: 1099-1108
View details for DOI 10.1007/978-3-030-00665-5_105
- An analysis of leg muscle stretch using 3D digital image correlation International Journal of Organizational and Collective Intelligence (IJOCI) 2017; 7 (3)
-
Multi stage classification and segmentation of brain tumor
IEEE. 2016: 1628-1632
View details for Web of Science ID 000388117501144
-
Hybrid Approach for Brain Tumor Detection and Classification in Magnetic Resonance Images
IEEE. 2015: 162-166
View details for Web of Science ID 000380390900035