Project Summary:

Throughout the project work to create the SAMEER-TU Database, Infrared image processing laboratory has been setup for Thermography applications in the field of Security system and Navigation. The project team has carried out a rigorous survey of restoring the visual and infrared degraded images in poor visibility condition followed by different fusion techniques.

Outcomes of the Project:

[1] Dataset Design: The research team designed a database named Society of Applied Microwave Electronics & Engineering Research-Tripura University (SAMEER-TU) Database, which consists of visual & its corresponding thermal images of human faces, natural scene images, vehicles & also videos of vehicles in low visibility/illumination conditions. The database consists of around 5640 natural scenes images in different weather and illumination conditions, out of which 1020 images are captured in foggy condition, 250 images in poor illumination condition and 4370 images in normal condition. The images in outdoor environment are mainly influenced by two factors-Weather effect and Illumination effect.

Sample Images of SAMEER-TU Dataset

Natural Scene Images and Their Corresponding Thermal Images from SAMEER-TU Database

Face Images and Their Corresponding Thermal Images of SAMEER-TU Dataset

[2] Developed Technique: The research team also implemented a technique for enhancing the quality of degraded visual image/frames in poor visibility condition. Ome output of the developed method are given below.

Human Resource Development:

Smt. Tannistha Paul Project Assistant M.Tech (CSE) B.E. (CSE)

Publications:

1. Kakali Das, Mrinal Kanti Bhowmik, Barin Kumar De, “Background Subtraction Algorithm For Moving Object Detection Using SAMEER-TU Dataset”, Proceedings of 4th International Conference on Soft computing for Problem Solving (‘SOCPROS’ 2014), published by Springer, Vol. 1, pp. 279-291, 2014, ISBN: 978-81-322-2216-3 (print), 978-81-322-2217-0 (online).

Implementing Institutes:

• Parent Institute: Tripura University (A Central University), Tripura, India.
• Collaborating Institute-I: Jadavpur University, Kolkata, India.
• Collaborating Institute-II: Agartala Government Medical College (AGMC) and Gobind Ballav Pant (GBP) Hospital, Tripura, India.

Project Summary:

The proposed project aims to develop an infrared imaging facility for detection and diagnosis of breast cancer and pain thereby creating an infrared image database of patients suffering from breast cancer and pain like spondylosis (e.g. Cervical: neck pain, Lumbar: back pain), Prolapse intervertebral disc (pain radiating to hands/legs), arthritis (joint pain), frozen shoulder (shoulder pain) etc. The specific objectives of the project are:

• Bio-Medical Infrared Image processing laboratory setup for thermography application.
• Disease diagnosis facility.
• Collection and creation on infrared medical imaging.
• Quality assessment of the infrared imaging.

Human Resource Development:

Dr. Usha Rani Gogoi, Junior Research Fellow (JRF), DBT Project Ph.D. in Computer Science and Engineering, Tripura University

Dr. Shawli Bardhan, Junior Research Fellow (JRF) and Senior Research Fellow (SRF), DBT Project Ph.D. in Computer Science and Engineering, Tripura University

Outcomes of the Project:

[1] Establishment of Bio-Medical Infrared Image Processing Laboratory:
The Bio-medical Infrared Image Processing Laboratory at Tripura University (A Central University) was established on 5th March, 2014, funded by Department of Biotechnology (DBT), Govt. of India and has an experience in thermography related to detection of breast cancer and pain in human body in the domain of Biomedical Engineering. For more details visit here

[2] Involvement of Major Equipment(s):
A highly sensitive infrared camera, FLIR T650sc of 640×480 IR Resolution with thermal sensitivity <30mK@30 degree centigrade has been installed at Biomedical Infrared image processing laboratory.

[3] Designing of Real-Time Datasets for Disease Diagnosis:
• DBT-TU-JU Breast Thermogram Dataset: The acquisition of breast thermograms dataset has been performed on a variety of patients. Most of the patients coming for breast thermography are proved to be either healthy or cancer free based on the mammography report. Here, based on the result of X-ray Mammography and FNAC, the breast thermograms are categorized as Healthy and Diseased. The Healthy Breast thermogram group contain the thermograms of all those patients whose mammography or FNAC result is either Healthy or Dense Breast. Similarly, the patient’s thermogram whose mammography result shows of either having “Benign cyst” or “Malignant tumor” is considered as Diseased Breast thermograms. For more details visit click here.

Sample Images of DBT-TU-JU Breast Thermogram Dataset

• IR (Infrared) Knee Joint Dataset: The acquisition of knee thermogram dataset of each patient are collected depends on the type of diseases (i.e., belongs to arthritis affected patient group) and number of joints affected. The area/joint of thermography of each patient is suggested by medical experts depending on the type of diseases. The clinical history and result of the clinical observations related to disease along with patient details are also analyzed during the dataset acquisition. For more details visit click here

IR (Infrared) Knee Joint Dataset

[4] Organization of Awareness Camp:
• Designing and Publication of Breast Cancer Awareness Booklet in BENGALI language.
• Organized Breast cancer Awareness Camp-1 and Camp-2 in collaboration with Mohana: A Women’s Association of Tripura University.

[5] Seminars/ Workshops Organized:
- Organized a workshop entitled “Computational Aspects of Inflammatory Pain Related Diseases” in Department of Computer Science & Engineering, Tripura University (A Central University) on 20th November, 2015.

Patent:

• Filed an Indian patent entitled “A system and method for segmenting suspicious hyperthermic regions from breast thermograms”, Application No: 202031027352, Published on: December 31, 2021.

Publications:

Total Publications:

• Journals: 11       • Conference Proceedings: 11       • Book Chapters: 02

Significant Publications:

1. Usha Rani Gogoi, Mrinal Kanti Bhowmik, and Gautam Majumdar, "MMSHRs: A Morphology Model of Suspicious Hyperthermic Regions for Degree of Severity Prediction from Breast Thermograms", Quantitative InfraRed Thermography, Taylor and Francis Online, Indexed by Science Citation Index Expanded (SCIE), pp. 1-25, 2022, Impact Factor: 1.643.[Citations: 01 as on 16/05/2024]


2. Shawli Bardhan, Satyabrata Nath, Tathagata Debnath, Debotosh Bhattacharjee and Mrinal Kanti Bhowmik, "Designing of an Inflammatory Knee Joint Thermogram Dataset for Arthritis Classification Using Deep Convolution Neural Network", Quantitative InfraRed Thermography Journal (QIRT), Taylor & Francis Online, Indexed by Science Citation Index (SCI), Vol. 19, No. 3, pp. 145-171, 2020, Impact Factor: 1.643. [Citations: 09 as on 16/05/2024]


3. Usha Rani Gogoi, Gautam Majumdar, Mrinal Kanti Bhowmik and Anjan Kumar Ghosh, "Evaluating the Efficiency of Infrared Breast Thermography for Early Breast Cancer Risk Prediction in Asymptomatic Population", Infrared Physics & Technology, Published by Elsevier, Indexed by Science Citation Index Expanded (SCIE), Vol. 99, pp. 201-211, 2019,Impact Factor: 3.3. [Citations: 48 as on 16/05/2024]


4. Usha Rani Gogoi, Mrinal Kanti Bhowmik, Debotosh Bhattacharjee, Anjan Kumar Ghosh, "Singular Value based Characterization and Analysis of Thermal Patches for Early Breast Abnormality Detection", Australasian Physical & Engineering Sciences in Medicine, published by Springer, Indexed by Science Citation Index Expanded (SCIE), Vol. 41, No. 4, pp. 861-879, 2018, Impact Factor: 4.4. [Citations: 41 as on 16/05/2024]


5. Sourav Pramanik, Debapriya Banik, Debotosh Bhattacharjee, Mita Nasipuri, Mrinal Kanti Bhowmik, and Gautam Majumdar "Suspicious-region segmentation from breast thermogram using DLPE-based level set method", IEEE Transactions on Medical Imaging, Indexed by Science Citation Index (SCI), Vol. 38, No. 2, pp. 572-584, 2018, Impact factor: 10.6. [Citations: 41 as on 16/05/2024]


6. Mrinal Kanti Bhowmik, Usha Rani Gogoi, Gautam Majumdar, Dhritiman Datta, Anjan Kumar Ghosh, Debotosh Bhattacharjee, "Designing of Ground Truth Annotated DBT-TU-JU Breast Thermogram Database towards Early Abnormality Prediction", IEEE Journal of Biomedical and Health Informatics (JBHI) (Formerly IEEE Transactions on Information Technology in Biomedicine),Indexed by Science Citation Index (SCI), Vol. 22, No. 4, pp. 1238-1249, 2018, Impact Factor: 7.7. [Citations: 48 as on 16/05/2024]

Implementing Institutes:

• Tripura University (A Central University), Tripura, India.

Project Summary:

The proposed project aim is to design an algorithm towards an automated decision support system for predicting some sort of abnormality such as inflammatory pain automatically and try to characterize the feasibility of thermal Imaging in diagnosis of Pyrexia of unknown origin. The specific objectives of the project are:

• Study and critical analysis of challenges of using infrared images for medical purpose.
• Develop an Algorithm for automatic abnormality prediction based on Infrared Images or other modality images (if exists) of collected & created Dataset.
• Performance Evaluation of the System & Comparative study with the Ground Truth data.

Human Resource Development:

Dr. Kakali Das, Senior Research Fellow (SRF), ICMR Project Ph.D. in Computer Science and Engineering, Tripura University

[1] Designing of Real-Time Datasets for Disease Diagnosis:
• Designing of Pain Thermogram Dataset: A benchmark dataset of pain thermogram are created under the project. This dataset consist of thermograms of various pain related disease including Cervical spondylitis, Arthritis (Gouty arthritis, Reactive arthritis, Rheumatoid arthritis, Mono arthritis, Osteoarthritis, Polyarthritis, Knee pain, and Seronegative Spondyloarthropathies. By following all the standard protocols for acquisition of thermograms related to pain diseases, a total of 80 thermograms are acquired covering different pain disease at age group 30-75. The statistics of the created dataset is shown in below Table 1. Some sample images are displayed in Fig. 1.

Thermogram Description	No of Thermograms	Age Group
Total no of thermogram captured :	80	30-75
a. No of pain disease affected thermogram :	76	30-75
1.Cervical spondylosis:	5	30-55
2. Arthritis :	71	35-75
i.Gouty arthritis:	5	50
ii. Reactive arthritis:	13	35-75
iii.Rheumatoid arthritis:	11	40-60
iv. Monoarthritis:	2	65-75
v. Osteoarthritis:	33	40-60
vi. Polyarthritis:	2	41
3. Knee pain:	3	38-80
4. Seronegative Spondyloarthropathies:	2	50-60

Sample Images of Created Pain Thermogram Dataset

[2] Designing of an Algorithm for automatic segmentation of affected region:
• Algorithm 1: The primary goal for the proposed segmentation algorithm is to fuse information from multiple segmentations of the input thermogram. The motivation being to find the core and stable hotspot region(s) quickly arriving at a consensus from multiple information sources. We propose to introduce information theoretic measure to achieve at consensus image segmentation. Variation of information (VoI) is the information theoretic measure used in our approach. VoI is used to fuse an ensemble of primary image segmentation techniques.

Block Diagram of the Proposed Algorithm-1

Example of segmentation ensemble and fusion result. First three rows of the image (15 segmentation maps) show Fuzzy-C means Clustering results. First image of the last row is the thermogram and the second image is the result of proposed method

• Algorithm 2: In this segmentation algorithm, the limitations with respect to the over and under segmentation of the inflamed regions is overcome by extending the conventional region growing segmentation technique with stronger similarity criteria and stopping rule. In this method, hotspot or inflamed region is generated by taking the intersection of two independent regions produced by two different version of Region growing algorithm using a separate set of parameters. An automatic multiseed selection procedure ensures prevention of missed segmentation

Sample segmentation maps using our proposed algorithm-2

• Algorithm 3: Thermal images of a human body part with any abnormality creates a persistent hotspot on that image. A region growing (RG) produces good accuracy in the segmentation of the hotspot from the thermal image. However, smooth edges in a thermal image leads to a decrease in accuracy of the hotspot segmentation procedure, as it may include an unwanted region around the edge or it may discard important areas around the edge. Therefore, performing an RG on the original image will not provide accurate segmentation results. We proposed an RG by modifying the stopping criteria and seed selection procedure. We employed the proposed RG on the original image as well as the log-transformed image. One low pass filter is used after a logarithmic transformation. An RG was performed on the filtered image. The output from this filtered image was not enough to produce an accurate segmentation. That is why we employed an RG on both the original image and filtered image.

Block Diagram of the Proposed Algorithm-3

Segmentation Results using Algorithm-3

Publications:

Total Publications:

• Journals: 03       • Conference Proceedings: 02      

Significant Publications:

1. Kakali Das, Satyabrata Nath, and Mrinal Kanti Bhowmik "Role of Infrared Imaging for Detection of Subclinical Inflammation of Joint", Journal of Clinical and Diagnostic Research, JCDR Pre Publishing, Indexed by Web of Science, Volume: 16, Issue: 2, pp: 8-12, 2022.


2. Kakali Das, Mrinal Kanti Bhowmik, and Dipti Prasad Mukherjee, "Segmentation of Knee Thermograms For Detecting Inflammation" Proceedings of 26th IEEE International Conference on Image Processing (ICIP) - Tier 2 Conference, pp. 1550-1554, 2019, Taipei, Taiwan. [Citations: 02 as on 24/05/2024]


3. Mrinal Kanti Bhowmik, Kakali Das and Debotosh Bhattacharjee, "Temperature Profile Guided Segmentation for Detection of Early Subclinical Inflammation in Arthritis Knee Joints From Thermal Images" in Infrared Physics & Technology, Published by Elsevier, Indexed by Science Citation Index Expanded (SCIE), Volume 99, pp 102-112, 2019, Impact Factor: 2.997. [Citations: 05 as on 24/05/2024]


4. Kakali Das, Mrinal Kanti Bhowmik, Omkar Chowdhuary, Debotosh Bhattacharjee and Barin Kumar De, "Accurate segmentation of inflammatory and abnormal regions using medical thermal imagery" in Physical and Engineering Science in Medicine (Formally Australasian Physical and Engineering Science in Medicine), Published by Springer, Indexed by Science Citation Index Expanded(SCIE), 2019, ISSN 0158-9938, Impact Factor: 7.099. [Citations: 04 as on 24/05/2024]

Implementing Institutes:

• Tripura University (A Central University), Tripura, India.

Human Resource Development:

Dr. Anu Singha, Senior Research Fellow (SRF), DRDO Project Ph.D. in Computer Science and Engineering, Tripura University

Dr. Rajib Debnath, Junior Research Fellow (JRF), DRDO Project Ph.D. in Computer Science and Engineering, Tripura University

Dr. Sourav Dey Roy, Junior Research Fellow (JRF), DRDO Project Ph.D. in Computer Science and Engineering, Tripura University

Shri Santanu Das, Studentship/ Assistantship, DRDO Project Ph.D. Pursuing in Computer Science and Engineering, Tripura University

Smt Puja Das, Studentship/ Assistantship, DRDO Project Ph.D. Pursuing in Computer Science and Engineering, Tripura University