ASAnanya Subramaniam

Ananya Subramaniam

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Ananya Subramaniam has a total of 28 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Opportunistic automated aortic valve calcification assessment in low-cost, screening CT calcium score exams

    by Ananya Subramaniam, Hao Wu, Sepideh Azarianpour, Naomi Joseph, Tao Hu, Neda Shafiabadi Hassani, Ammar Hoori, Sanjay Rajagopalan, Sadeer Al-Kindi, David L. Wilson

    Aortic valve stenosis (AS) is the most common valvular disease, with a growing impact in the aging population. AS can culminate in heart failure if left untreated. Early treatment with minimally-invasive transcatheter aortic valve replacement (TAVR) is being evaluated in clinical trials. We addressed an unmet clinical need for early detection, referral to echocardiography for AS evaluation, and possible treatments (e.g., lifestyle changes, drugs, or TAVR). As aortic valve calcification (AVC) is typically present in AS, we created a method to detect AVC in low-cost/no-cost non-contrast CT calcium score (CTCS) screening exam images. We developed a multi-task deep network to identify a cylindrical aortic valve region of interest (ROI) and applied Agatston criteria within the ROI to obtain calcifications. Predicted ROIs had good agreement with cardiologists’ labels and sometimes were better, and predicted Agatston scores agreed with cardiologists (r = 1.00, paired t-test p = 0.573, t = 0.57). On a retrospective screening cohort of 2000 + patients, we found that 20.6% of men and 22.2% of women had some degree of AVC. According to guidelines, 3.53% and 9.04%, respectively, would have severe AS. These promising results indicate that further evaluation of this approach is warranted, with the potential for significant public health impact.

  • Improving coronary artery segmentation with self-supervised learning and automated pericoronary adipose tissue segmentation: a multi-institutional study on coronary computed tomography angiography images

    by Justin N. Kim, Yingnan Song, Hao Wu, Ananya Subramaniam, Jihye Lee, Mohamed H. E. Makhlouf, Neda S. Hassani, Sadeer Al-Kindi, David L. Wilson, Juhwan Lee

    PurposeCoronary artery disease (CAD) is a leading cause of morbidity and mortality worldwide, with coronary computed tomography angiography (CCTA) playing a crucial role in its diagnosis. The mean Hounsfield unit (HU) of pericoronary adipose tissue (PCAT) is linked to cardiovascular risk. We utilized a self-supervised learning framework (SSL) to improve the accuracy and generalizability of coronary artery segmentation on CCTA volumes while addressing the limitations of small-annotated datasets.ApproachWe utilized self-supervised pretraining followed by supervised fine-tuning to segment coronary arteries. To evaluate the data efficiency of SSL, we varied the number of CCTA volumes used during pretraining. In addition, we developed an automated PCAT segmentation algorithm utilizing centerline extraction, spatial-geometric coronary identification, and landmark detection. We evaluated our method on a multi-institutional dataset by assessing coronary artery and PCAT segmentation accuracy via Dice scores and comparing mean PCAT HU values with the ground truth.ResultsOur approach significantly improved coronary artery segmentation, achieving Dice scores up to 0.787 after self-supervised pretraining. The automated PCAT segmentation achieved near-perfect performance, with R-squared values of 0.9998 for both the left anterior descending artery and the right coronary artery indicating excellent agreement between predicted and actual mean PCAT HU values. Self-supervised pretraining notably enhanced model generalizability on external datasets, improving overall segmentation accuracy.ConclusionsWe demonstrate the potential of SSL to advance CCTA image analysis, enabling more accurate CAD diagnostics. Our findings highlight the robustness of SSL for automated coronary artery and PCAT segmentation, offering promising advancements in cardiovascular care.


  • Open Access

    Image harmonization and deep learning automated classification of plus disease in retinopathy of prematurity

    by Ananya Subramaniam, Faruk Orge, Michael Douglass, Basak Can, Guillermo Monteoliva, Evelin Fried, Vanina Schbib, Gabriela Saidman, Brenda Peña, Soledad Ulacia, et al.

    Retinopathy of prematurity (ROP), a disease affecting preterm infants characterized by abnormal development of retinal blood vessels, is the leading preventable cause of infantile blindness.1,2 Plus disease is an indicator of ROP severity and is characterized by severe dilation and tortuosity of retinal blood vessels in the posterior pole.1,3,4 Once plus disease is identified, infants are treated using laser photocoagulation or anti-vascular endothelial growth factor (anti-VEGF) injections with great success.5,6 However, an incorrect diagnosis or lack of screening can prove detrimental as the onset of plus disease can result in blindness within two weeks if left untreated, with nearly 50,000 infants blinded due to ROP annually worldwide.7,8 Routine retinal examination by skilled ophthalmologists is critical for early detection and treatment of ROP.