Oral Session 2B

December 21,   1:30 PM to 2:30 PM

Chair: Venkatesh Babu Radhakrishnan

          
67 Interpretive Self-Supervised Pre-training: Boosting Performance on Visual Medical Data
December 21,   13:30:00 to 13:45:00
Authors: Siladittya Manna (Indian Statistical Institute, Kolkata)*; Saumik Bhattacharya (Indian Institute of Technology Kharagpur); Umapada Pal (Indian Statistical Institute, Kolkata)
Abstract: Self-supervised learning algorithms have become one of the best tools for unsupervised representation learning. Although self- supervised algorithms have achieved state-of-the-art performance for classification tasks in the case of natural image data, their application on medical data has been limited. In this work, we have proposed a novel loss function and derive it's asymptotic lower bound. We have also shown that self-supervised pre-training with the proposed loss function helps in surpassing the supervised baseline on the downstream task. We have also shown that the self-supervised pre-training helps a model in learning better representation in general to achieve better performance compared to supervised baselines. We have mathematically derived that the contrastive loss function asymptotically treats each sample as a separate class and works by maximizing the distance between any two samples and this helps to get better performance. Finally, through exhaustive experiments, we demonstrate that self-supervised pre-training helps to surpass the performance of fully supervised models on downstream tasks.
Presenting Author: Siladittya Manna
Paper: https://doi.org/10.1145/3490035.3490273
Joining link to attend this talk
December 21,   13:30:00 to 13:45:00
 
100 Multi-task learning based approach for surgical video desmoking
December 21,   13:45:00 to 14:00:00
Authors: Vartika Sengar (TCS Research)*; Vivek B S (TCS); Karthik Seemakurthy (TCS Innovation labs); Jayavardhana Gubbi (TCS Research); Balamuralidhar P ( Tata Consultancy Services)
Abstract: Over the past few decades, minimally invasive surgical techniques have gained wide acceptance due to multiple benefits it offers. In these surgeries, a camera with a light source is inserted via a small incision. The video feed from the camera is the only source for visualization of internal organs. Certain procedures produce fumes that severely degrade the video feed. Various image processing based de-smoking systems are proposed to provide a continuous, good quality video feed. However, most of the existing approaches perform de-smoking at the frame level and fail to exploit the dynamic properties of the smoke. We propose a novel de-smoking model that harnesses both spatial and temporal properties of smoke. We evaluate the performance of the proposed model on the Cholec-80 dataset and observe a superior performance in terms of MS-SSIM and PSNR metrics compared to existing works.
Presenting Author: Vartika Sengar
Paper: https://doi.org/10.1145/3490035.3490283
Joining link to attend this talk
December 21,   13:45:00 to 14:00:00
 
136 Magnetic resonance image reconstruction by nullspace based finite rate of innovation framework
December 21,   14:00:00 to 14:15:00
Authors: Sudhakar Reddy (National Institute of Technology Karnataka)*; B.S Raghavendra (National Institute of Technology Karnataka); A V Narasimhadhan (National Institute of Technology Karnataka)
Abstract: The finite rate of innovation (FRI) framework has been usedto reconstruct analog signals which have a finite numberof parameters. The FRI framework also finds its use in reconstructingimages from undersampled magnetic resonance(MR) data. Reconstructing MR image from MR data is anestimation problem which can be solved utilizing Pronysmethod. However, Pronys method involves solving polynomialroots of annihilating filters which leads to an unstablereconstruction in the high noise scenario. In this paper, weintroduce a novel reconstruction approach based on the annihilatingfilter that involves the utilization of solutions of anunderdetermined linear system. The proposed reconstructionthe approach shows superior peak signal to noise ratio (PSNR)and structural similarity index measure (SSIM) than that ofconventional FRI methods in the high noise-case simulationscenario.
Presenting Author: Sudhakar Reddy
Paper: https://doi.org/10.1145/3490035.3490294
Joining link to attend this talk
December 21,   14:00:00 to 14:15:00
 
191 Fusion of Image Quality Assessment and Transfer Learning for COVID19 Detection Using CT Scan Image
December 21,   14:15:00 to 14:30:00
Authors: Kiruthika S (IIITDM Kancheepuram)*; Masilamani V (IIITDM Kancheepuram); Joshi Pratik (IIITDM Kancheepuram)
Abstract: One of the main challenges in controlling the spread of COVID19 pandemic is to diagnose infection early. The most reliable method 𝑅𝑇 − 𝑃𝐶𝑅 takes several hours to give results. Although the Anti-Body (Serological) test gives the results in a few hours, it is not accurate, and hence it is not reliable. Moreover, they are invasive. Another issue with these methods is that the number of labs performing these tests are very limited. It will be beneficial if the already existing clinical infrastructure is used for diagnosing COVID19 accurately in real time. Recently chest CT images are used by researchers to diagnose the COVID19 with impressive accuracy. The state of the art method for detecting COVID19 using CT chest images involves Deep Learning. Deep Learning is expected to provide accurate and reliable results only when the model is trained on a large data set. Due to non-availability of a large data set the existing models have been trained on a smaller size data set. Therefore it would be better to design a model to give good accuracy with reliability. To achieve accuracy along with reliability we proposed a COVID19 detection model with the combination of deep learning model and the traditional machine learning model. The novelty of the proposed model is the fusion of image quality and deep learning. The proposed method outperformed the state of the art method in terms of accuracy, recall and F1 score (more than 99 % in almost all the metrics) on a benchmark data set. The efficacy of the selected features and also explainability of the method are demonstrated through various tests.
Presenting Author: Kiruthika S
Paper: https://doi.org/10.1145/3490035.3490307
Joining link to attend this talk
December 21,   14:15:00 to 14:30:00

    
December 20December 21December 22
Session 1A Session 2A Session 3A
Session 1B Session 2B Session 3B
Session P1 Session P2 Vision India
Plenary 1 Plenary 3 Plenary 4
Plenary 2    
List of Accepted Papers
Conference Program