Clinical Data Analytics & Radiomics

BII_Research-CIID-CDAR-2023

Research

The group’s vision and aim are development of novel clinical data analysis algorithms to extract meaningful information which enables detection, characterization, and quantification of clinically significant information. Our group aims to:

  • Develop new skills & knowledge in advanced multi-modal image (MR, CT, and PET) data processing,
  • Contribute to local research activities,
  • Forge collaborations with clinical community and industries
  • Develop new IPs, publish in prestigious journals and secure grants.

The group’s current capabilities are in Quantitative Biology and Radiomics

BII_Research-CIID-CDAR_figure-1

Figure 1. Our group focuses on developing Computer-aided Decision support systems for preclinical and clinical applications. The following paragraphs describe our efforts in this direction with use cases.

1. AI-augmented clinical radiology platform for detection of brain aging using Normative MRI Brain Aging Atlas, based on the local population in Singapore

Brain ageing differs from chronological ageing and reflects an individual’s brain health. Unlike chronological age, brain age can change, and accelerated ageing can be reversed with targeted therapy. Currently, the lack of AI-augmented platforms for quantifying advanced brain structural changes forces radiologists to rely on visual inspection, a subjective method that challenges even neuroradiologists.

An AI-augmented brain volumetry-based brain age detection platform (‘SaMD’),  is developed using normative population data in Singapore to build SG-Brain Template. This template will help to diagnose accelerated brain aging and facilitates differentiation from normal brain ageing. Many brain ageing studies were available for the Caucasian population. Owing to the difference in normative aging between the Caucasian and the Singapore population, there is a need to develop a normative atlas to detect accelerated brain aging in Singapore. It will serve as a local reference for normative brain volumetry leading to early detection, early intervention, and effective management. 

BII_Research-CIID-CDAR_figure-2A

Figure 2a. Data processing pipeline to generate brain atlas templates.

BII_Research-CIID-CDAR_figure-2B

Figure 2b. Atlas templates of different age groups developed.

2. AI/ML based identification of Neuroanatomical markers for Schizophrenics and subtype classification

"Schizophrenia is a major psychiatric disorder with significant clinical burden on patients and caregivers. Identifying classification biomarkers can enhance clinical measures and our understanding of schizophrenia's neural basis. Previous machine learning studies using neuroanatomical features have attempted to classify schizophrenia from healthy controls but have used a limited range of measures. This study classified schizophrenia and healthy controls using diverse neuroanatomical measures (cortical and subcortical volumes, cortical areas and thickness, cortical mean curvature) and Ensemble methods for better performance. We also correlated these features with Quality of Life (QoL) scores in the schizophrenia cohort. Ensemble methods and diverse measures achieved classification accuracies of 83-87%, sensitivities of 90-98%, and specificities of 65-70%. Lower QoL scores in the schizophrenia cohort were significantly correlated with specific neuroanatomical measures and psychological health and social relationship domains of QoL. Our results highlight the value of diverse neuroanatomical measures and Ensemble methods for better classification performance and understanding the neurobiological correlates of QoL in schizophrenia." 

In this study, we are working on

  1. Quantitative Neuroimaging Features for SCZ /HC
  2. Correlation studies – features vs underlying conditions, features vs psychometric scores etc.
  3. Neuroimaging-based condition Detection/ Staging
  4. Neuroanatomy subtypes and their relation to the management of the condition
BII_Research-CIID-CDAR_figure-3

Figure 3. Neuroanatomical subtypes of Schizophrenia and their correlation to PANSS and QOL.

3. Comprehensive Abdominal Fat Analysis Tool for Geriatric Cohort Study

With the increasing prevalence of obesity and its tremendous health effects, it is important to invest time and effort into research of obesity and its associated diseases to develop better care and prevention. Worldwide, there are many clinical trials like – GENYAL (prevention of obesity in childhood); SWITCH (Community-based obesity prevention trial); a Clinical trial exploring exercise intervention on obese women; SAMS (Singapore adult Metabolism Study); GUSTO (Growing towards a healthy outcome in Singapore); etc. and many countries like Singapore have declared war on obesity. Understanding the phenotypes and genotypes of obesity is crucial for profiling and management of the condition. In-vivo quantification of obesity aids in monitoring changes due to interventions, visualizing fat compartments, examining inter-ethnicity differences, and tracking inter- and intra-subject variations in large cohort studies. In large cohort studies, there is a data deluge, and we need a comprehensive tool which includes (i) Accurate, rapid, and repeatable segmentation module (ii) Module for visualization of data and statistics and (iii) A correction tool to correct for errors if any.

BII_Research-CIID-CDAR_figure-4

Figure 4. End to End Human Metabolic Fat Analysis Framework.

In this study, we propose “CAFT: Comprehensive Abdominal fat segmentation tool” which comprises of a deep learning framework “Global aggregation UNet with self-attention (GA-UNet)” to accurately segment all the three compartments SSAT, DSAT and VAT simultaneously using a single network architecture. Dashboard for data presentation and analytics – which allows automatic lumbar based quantification and analysis; 3D visualization; Percentage and volumetric analysis of whole abdomen or lumbar based fat and an Interactive correction tool to edit the contours (between background & outer abdominal boundary; between SSAT & DSAT – correction of Fascia Superficialis and inner abdominal boundary which separates SAT and VAT) to correct the errors in segmentation.

4. Automated Analysis of Cardiac MRI for Clinical Trials

The study aims to speed up segmentation and improve interobserver agreement in CMR imaging analysis. Automated machine learning shows promise in eliminating variation. We aim to develop an AI algorithm for LV segmentation and other CMR aspects like infarct size, T1, T2, and LGE analysis. To reduce radiologists' processing time and variability, we propose a fully automated multi-scan CMR image analysis pipeline. This pipeline aims to match human expert precision and achieve global standardization using multi-centre trial data from IMMACULATE/PITRI/Platelet-STEMI cohorts. The pipeline has significant translational potential and will advance ML and AI research. Plans include applying it to the Singapore Biobank, licensing it to CVI42 and other vendors, and offering a superior CMR Core lab service to international trial groups.

Methods: Sequence-specific U-Net 2D models were trained for segmenting the left ventricle (LV), right ventricle (RV), and aorta in cine short-axis and late gadolinium enhancement sequences as needed. The models were trained and tested on data manually segmented by experts using semi-automatic and manual tools. Parameters such as the left ventricle ejection fraction (EF) and scar percentage were computed from the segmentations. The Dice similarity coefficient, Hausdorff distance, mean surface distance, and Pearson correlation coefficient R were used to assess and compare the U-Net-based pipeline results with intra-observer variability.

5. Brown Fat and Metabolic Health

All fats in the body are not similar and have different functions.  The body has different fat types. White fat stores excess energy, while brown fat dissipates it as heat through thermogenesis. Traditionally, brown adipose tissue (BAT) was thought to be abundant only in infants and to regress with age. Recent studies show BAT persists into adulthood but is less prevalent in obese individuals. BAT is an appealing target for combating obesity and metabolic diseases like diabetes due to its ability to use fat stores as fuel.

We investigated the influence of brown fat on adiposity in 198 young Asian preschool children from the GUSTO study using MRI, a radiation-free technique to identify brown fat. Our study is the first to examine brown fat concerning metabolic health in children from a general population. Current methods for identifying brown fat, like PET/CT and infrared thermography, detect only activated brown fat. Our findings confirm that MRI can identify and quantify brown fat without activation, paving the way for future studies in identifying brown fat in healthy populations and in longitudinal studies without radiation.

Members

 Principal Scientist PRAKASH K.N. Bhanu   |    [View Bio]  
 Scientist CHILLA V.N. Geetha Soujanya 
 Lead Research Officer C.S. Arvind 

Selected Publications

  1. Bhanu K.N. Prakash, Arvind Channarayapatna Srinivasa, Ling Yun Yeow, Wen Xiang Chen, Audrey Jing Ping Yeo, Wee Shiong Lim and Cher Heng Tan. MultiRes Attention Deep Learning Approach for Abdominal Fat Compartment Segmentation and Quantification. Publication: IntechOpen (2023). https://doi.org/10.5772/intechopen.111555

  2. Bhanu Prakash KN, Arvind CS, Abdalla Mohammed, Krishna Kanth Chitta, Xuan Vinh To, Hussein Srour & Fatima Nasrallah. An end-end deep learning framework for lesion segmentation on multi-contrast MR images—an exploratory study in a rat model of traumatic brain injury. Publication: Springer, Medical & Biological Engineering & Computing (2023). https://doi.org/10.1007/s11517-022-02752-4

  3. Yeow, Ling Yun; Teh, Yu Xuan; Lu, Xinyu; Srinivasa, Arvind Channarayapatna; Tan, Eelin ; Tan, Timothy Shao Ern; Tang, Phua Hwee; KN, Bhanu Prakash .  Prediction of MYCN Gene Amplification in Pediatric Neuroblastomas: Development of a Deep Learning-Based Tool for Automatic Tumor Segmentation and Comparative Analysis of Computed Tomography-Based Radiomics Features Harmonization. Publication: Journal of Computer-Assisted Tomography (2023). https://doi.org/10.1097/RCT.0000000000001480

  4. Arvind Channarayapatna Srinivasa, Ling Yun Yeow, Wen Xiang Chen, Audrey Jing Ping Yeo, Wee Shiong Lim and Cher Heng Tan, Bhanu K.N. Prakash. Comprehensive Adipose Tissue/Muscle Segmentation and Quantification Tool for Sarcopenia Prediction, SGCR-WIRES 2023

  5. Bhanu Prakash KN, C.S Arvind, S.Z.T. Jordan, L.J. Chew, J.P.A. Yeo, W.S.Lim, C.H.Tan Singapore/SG European Society of Radiology AI Framework for Quantifying Muscular and Adipose Tissue Variation in the Elderly: An MRI-Based Study of a Geriatric Singapore Cohort. (ECR - 2024)

  6. Ling Yun Yeow, Chi Long Ho, Tan Xu Hao Isaac, Parag R. Salkade, Oliver James Nickalls, Shea Foo, Bhanu Prakash KN.  Comparative analysis of normative brain structures and volumes between Singapore (SG) Chinese and Caucasian (Cau) populations. ECR 2024

  7. KN Bhanu Prakash, Arvind Channarayapatna Srinivasa, Dikendra Baduwal A. Hegde, O. Nickalls, P. Salkade, C. L. Ho; NeuroAI: Constructing an SG- Population-specific Normative Brain Ageing Volumetry Database for Advanced Cognitive Health Assessment. ECR 2024.

  8. Arvind Channarayapatna Srinivasa, Seema Bhat, Dikendra Baduwal, Sim Zheng Ting Jordan Ashwin Amarapur, KN Bhanu Prakash. Enhancing quick-acquried MRI scans with the DL-based Aikenist framework: a clinical assessment. European Society of Radiology (ECR – 2024).

  9. Jordan, Arvind Channarayapatna Srinivasa, Justin Audrey Jing Ping Yeo, Wee Shiong Lim and Cher Heng Tan, Bhanu K.N. Prakash. Automated Comprehensive Fat Quantification and Sarcopenia Assessment Tool - A Potential Magnetic Resonance Imaging Biomarker. Abdominal Radiology Group of Australia and New Zealand (ARGANZ), Melbourne, Aus 23-2-2024 

  10. Ling Yun, Arvind Channarayapatna; Tan, Eelin; Tan, Timothy Shao Ern; Tang, Phua Hwee; KN, Bhanu Prakash Deep Learning-based tumor segmentation and radiomic features-based classification of MYCN gene amplification status in pediatric neuroblastoma. SGCR-WIRES 2022

  11. Ling Yun Yeow, Chi Long Ho, Tan Xu Hao Isaac, Parag R. Salkade, Oliver James Nickalls, Shea Foo, Bhanu Prakash KN. Comparative analysis of normative brain structures and volumes between SG and Caucasian population. AOCNR-SGCR-WIRES 2023

  12. Ling Yun Yeow, Chi Long Ho, Tan Xu Hao Isaac, Parag R. Salkade, Oliver James Nickalls, Shea Foo, Bhanu Prakash KN.  Machine Learning Approach to Brain Age Prediction (BAP) for Local Population in Singapore. AOCNR-SGCR-WIRES 2023.

  13. Ling Yun Yeow, Chi Long Ho, Tan Xu Hao Isaac, Parag R. Salkade, Oliver James Nickalls, Shea Foo, Bhanu Prakash KN.  Age-specific MRI brain templates/atlases for healthy Chinese population in Singapore. AOCNR-SGCR-WIRES 2023.

  14. Chew, Q. H., Bhanu Prakash KN., Koh, L. Y., Chilla, G., Yeow, L. Y., & Sim, K. (2022). Neuroanatomical subtypes of schizophrenia and relationship with illness duration and deficit status. Schizophrenia Research, 248, 107-113. (Chew & Bhanu Prakash are joint first authors).

  15. Geetha Soujanya Chilla, Yeow Ling Yun; Qian Hui Chew, Kang Sim, Bhanu Prakash KN. "Machine learning classification of schizophrenia patients and healthy controls using diverse neuroanatomical markers and Ensemble methods" Sci Rep 12, 2755 (2022). https://doi.org/10.1038/s41598-022-06651-4. [IF: 5.035]