Malay SINGH
Projects
- mIHC Prostate Highly Mulitplexed Immuno-fluorescence Image Data Analysis for Prostate Cancer
We have developed an automated image processing pipeline to detect and quantify multiple immunophenotypes in histological prostate tissue samples. Our pipeline integrates powerful image processing and machine learning methods to analyse histopathological tissue whole slice images (WSI). More details of the project are here.
- Cribriform Pattern Detection
- June 2017 - Oct 2019
Architecture, size, and shape of glands are most important patterns used by pathologists for assessment of cancer malignancy in prostate tissue slides. Varying structures of glands along with cumbersome manual observations may result in inaccurate assessment. Cribriform gland with irregular border is an important feature in Gleason pattern 4. We are developing a deep learning based cribriform pattern classification system, which uses several pre-trained deep learning models that were fine-tuned with our own labelled H&E image dataset. These fine-tuned deep learning models provide us pre-defined features derived from natural images which are transferred to histopathological imaging domain. We have achieved promising results with these fine-tuned deep learning models.
ArXiv Pre-print (October, 2019) PDF. - Nuclear Pleomorphism in Renal Clear Cell Cancer
- June 2015 - December 2017
The characteristics of the nuclei are often observed by pathologists when they assess the progression and presence of cancer cells in tissue biopsies. Cancerous tissue typically contains cells with enlarged, irregularly-shaped (pleomorphic) and darkly-stained (hyperchromasia) nuclei with prominent nucleoli. However, at different stages of the disease, the nuclear structure and prominence of nucleoli can change. The Fuhrman grading system for clear cell Renal Cell Carcinoma (ccRCC) was developed around these observed changes in the nuclei. It provides rules to classify the different stages of disease progression. Early stage ccRCC tumors typically have small, round nuclei with inconspicuous nucleoli, while late stage tumors have enlarged and irregularly-shaped nuclei with prominent nucleoli. Following on from our work on nucleoli detection, we have developed new machine learning methodologies to perform automatic grading of ccRCC histopathological images. From the histopathological images, we extract features describing the properties of multiple nuclei concurrently. This enables us to train classifiers that can distinguish the level of pleomorphism of the nuclei in the tissue sample, resulting in a higher accuracy in the automated grading.
Published in the JCO Clinical Cancer Informatics (April, 2018) PDF. - Automated Image Based Tumor Risk Assessment System for Hepatocellular Carcinoma
- June 2016 - December 2017
The evaluation of both asymptomatic patients and those with symptoms of liver disease involves blood testing and imaging evaluation. We developed an automated image based tumor risk assessment system as part of a micro-array gene expression based prognostic stratification system for resectable hepatocellular carcinoma. Whole slide images of liver cancer tissue were divided into two groups namely "Low Risk" and "High Risk" by micro-array gene expression based prognostic stratification system. These slides were then immunohistochemically (IHC) stained for different biomarker proteins. We developed an automated image based system to analyse the biomarker protein content. Our system predicted a Support Vector Regression (SVR) based score for each IHC image after quantification and analysis of stain. Our system was able to predict a higher SVR score for "High risk" patients when compared to "Low Risk" patients.
Published in Molecular Oncology Journal (December, 2017) PDF. - Gland Segmentation in Prostate Histopathological Images
- June 2015 - June 2017
Glandular structural features are important for the tumor pathologist in the assessment of cancer malignancy of prostate tissue slides. The varying shape and size of glands combined with tedious manual task can result in inaccurate assessment. There are also discrepancies and low level agreement among pathologists especially in cases of Gleason pattern 3 and pattern 4 prostate adenocarcinoma. We have developed an intelligent software to improve accuracy and reduce labor of gland structure assessment on Haematoxylin and Eosin (H&E) stained prostate tissue slides. Our method can easily fit into the existing workflow of the pathologist. Prostate cancer glands with their varying shapes, structures, and size pose an extreme challenge for automated gland segmentation systems. Our method achieved an averaged Jaccard Index score of 0.54 (range is [0,1], higher value is better) while outperforming various existing softwares in the literature.
Published in the Journal of Medical Imaging (June, 2017) (JMI). - Automated Image Based Prominent Nucleoli Detection
- May 2013 - June 2015
The diagnosis and prognosis of cancers are major issue for a trained pathologist. Inter-observer variability and tediousness of tissue reading hamper the accuracy of assessment by the pathologist. The analysis of prominent nucleoli is one of the main methods of cancer assessment. We have developed an intelligent software to improve accuracy and reduce labor of tissue reading of prominent nucleoli assessment on H&E stained slides. Our method can easily fit into the existing workflow of the pathologists work.
Published in the Journal of Pathology Informatics. (June, 2015) JPI.
An extended study of nuclei classification methods was also presented at International Conference on Innovation in Medicine and Healthcare, (KES-InMed-17), (SpringerLink).
- Speech Retrieval
- January 2012 - July 2012
The Project presents an approach to multi-modal processing in speech technology. We suggested and implemented the idea of how a Speech corpus containing large amount of audio data can be transcribed offline and used on-the fly by a simple text index. We proposed a novel approach to corpus storage in groups which would help in turn to categorize the data stored. During retrieval process the groups to be searched may be defined.
Presented in IMPACT 2013. IEEEXplore - Human Emotion Recognition
- July 2011 - December 2011
In this project, we made a emotion recognition software. The software was trained over a dataset collected by survey of IIITA Students. We asked to tag a certain Hindi Song for emotion the songs as it depicts to them. The four emotions are Happy,Sad,Anger,Calm. We classified a new song using Artificial Neural Network and K- Nearest Neighbor Implementation in C++. The song corpus consisted the Hindi Songs from 1950's.
- Optimization of CO2 Injectivity in Geological Carbon Storage
- May 2011 - July 2011
In this project, we implemented a model of Oil reservoirs. We tried to infer a relation between Well location and migration of CO2 in depleted or near-depleted Oil Reservoirs.We were to optimize the amount of Carbon Dioxide injected into the reservoir within the constraints of Parameters like Bottom Hole Pressure, Permeability and Porosity of Rocks.
- A Cognitive Interactive Framework for Multi-document Summarizer
- January 2011 - May 2011
In this project, we made a generic interactive framework based on human cognition, where the system can learn continuously from the Internet and from its interaction with the users. To show the utilization of this framework, Iintelli, an agent based application for multiple text document summarization was developed and compared with the MEAD on the Cran Data Set. The human knowledge and experience were represented through fuzzy logic-based word-mesh and sentence-mesh which can learn. Learning was performed using the competitive models Maxnet and Mexican Hat.
Presented in IHCI 2011. SpringerLink - K - Coverage in Wireless Sensor Networks
- July 2010 - November 2010
For a given area A with randomly placed wireless sensor nodes, we designed an algorithm to schedule the sensors so as to maximize the network lifetime while at the same time ensuring that subsets of the given area (A1,A2,.., An) are(k1,k2,..,kn )-covered respectively. Probabilistic model was implemented to improve upon network lifetime for sensor networks.