Portfolio

About memy stats

My Skills

Machine Learning

92%

Deep Learning

80%

NLP

70%

Python

90%

95%

Java

80%

C++

75%

HTML, CSS

85%

JavaScript

70%

SQL, PSQL

77%

PHP

60%

Internet of Things

65%

My Publications

Springer · March 14, 2026

CoMFDNet: a correlation-guided meta-fusion based dual network for glaucoma classification
- Ananya Saha, Gouranga Maity, Ram Sarkar

Glaucoma is a serious eye disease that causes permanent blindness if not detected early. Since the disease develops slowly and the changes in the optic nerve are very subtle, early diagnosis from retinal fundus images is challenging. In this paper, we propose a new method called Correlation-guided Meta-Fusion based Dual Network (CoMFDNet) for automatic glaucoma classification. Our model uses two feature extractors: ResNet50, which captures deep structural and semantic features of the optic nerve head, and MobileNetV2, which provides lightweight texture-level features for faster processing. To combine these features, we have used a correlation-based fusion module that measures similarity between features using cosine similarity. This helps the model keep useful and complementary features while reducing redundant information. We also use Efficient Channel Attention (ECA) and Spatial Attention to highlight the most important retinal regions, such as the optic disc and cup, which are key in glaucoma diagnosis. Finally, a meta-learner adaptively adjusts the contribution of each branch, making the model more flexible and robust. We have tested our model on two standard glaucoma datasets. On the ACRIMA dataset, CoMFDNet achieves an accuracy of 99.63% and an F1-score of 99.62%. On the LAG dataset, it achieves an accuracy of 97.32% and an F1-score of 97.08%. The results show that our approach outperforms existing methods while remaining efficient for real-world use in clinical screening.

Elsevier · Feb 12, 2025

NeuralCodOpt: Codon optimization for the development of DNA vaccines
- Tapan Chowdhury, Aishwarya Saha, Ananya Saha, Arnab Chakraborty, Nibir Das

Inefficient gene translation, driven by organisms’ codon preferences, is an emerging research area since this results in sluggish processes and diminished protein yields. Our research culminates in deriving efficient, optimized codon sequences by considering organism-specific Relative Codon Adaptiveness (RCA) ranges. In this research work, we have developed a novel algorithm, Neural Codon Optimization (NeuralCodOpt), to automate the process of codon optimization tailored to a specific organism and input sequence. Our algorithm has two main parts: the target Codon Adaptation Index generation using K-Means and the automation of sequence optimization using reinforcement learning. This algorithm has been tested across a set of 130 species, yielding highly optimal results that are quite significant compared to the previous works. NeuralCodOpt has shown a high accuracy of 86.7%, which would substantially contribute to Deoxyribonucleic Acid (DNA) vaccines by improving the efficiency of DNA expression vectors. These vectors are crucial in DNA vaccination and gene therapy as they enhance protein expression levels. By further incorporating it into plasmid construction, the translational efficiency of DNA vaccines will be significantly improved.

Invention Journals · May 5, 2023

Commercial aspect of Telemedicine in Indian scenario
- Prof. Jafor Ali Akhan, Ananya Saha, Dibyendu Rana, Dr. Himadri Nath Saha

Telemedicine (also referred to as "telehealth" or "e-medicine") was first introduced in the 1970s, when tele-electrocardiograms were communicated over telephone lines. Since then, telemedicine has come a long way in terms of both healthcare delivery and technology. It involves using electronic information to distribute health-related services when distance separates the participants. The Indian Space Research Organization has been deploying a SATCOM-based telemedicine network nationwide since 1999. During the COVID-19 pandemic, a countrywide lockdown in India reduced access to regular healthcare services. As a policy response, the Ministry of Health and Family Welfare rapidly increased the use of telemedicine. And now, India, with its diverse landmass, huge population, and large medical and IT manpower, has been found to be an ideal setting for telemedicine. For Low- or Middle-Income Country (LMIC), which includes India, investing in telemedicine is proven to be highly beneficial, making healthcare more accessible and unbiased in the future. Surely telemedicine cannot be a solution to all problems, but it can still reduce the burden on the healthcare system to a large extent.

Invention Journals · Mar 19, 2023

Heart Disease Treatment in Telemedicine using Internet of Things
- Prof. Jafor Ali Akhan, Ananya Saha, Shubhankar Ghosh, Dr. Himadri Nath Saha

Due to the unfold of infectious diseases, like COVID-19, and flu, and speedy virtualized services, telemedicine has received sizeable popularity. As a consequence of huge expand in smart devices and availability of internet facility, Internet of Things has massive demand for saving money on human labor, and providing security. With the growing needs for health requirements, telemedicine in the Internet of Things guarantees to be transformational due to the fact that physicians can track patients' health more effectively. We have proposed an innovative model for adoption to enforce a heart ailment detecting device for a telemedicine system. Our objective is to record and detect the heartbeat of the patient to screen the risk of a heart attack alongside with ordinary check-ups. The design of this low-cost and Internet of Things enabled fingertip-based critical heart disease detecting device makes our model innovative in contrast to existing models. Besides that, the proposed model provides feedback and notifications to the patient to control and mitigate various health complications. The heart rate module picks up heart rate signals from the patients and sends them wirelessly to a computer or Android application for visualization.