PILLAR  Lab- Computational Pathology

Identifying which patients with low-grade lymphoma are likely to undergo histological transformation is critical because it marks a shift to a significantly more aggressive disease state that requires a prompt transition from "watchful waiting" or mild therapy to intensive, high-dose chemotherapy regimens. We are exploring how computational pathology can assist in the prediction of low grade lymphoma transformation

Highlighting blasts on H&E sections is essential because their quantification is the gold standard for distinguishing between low-grade myelodysplastic syndromes and acute myeloid leukemia, yet this remains clinically problematic because blasts frequently lack distinguishing features. Using computer vision tools we are exploring novel ways to improve blasts identification

Correctly predicting which patients with early-stage (Stage I/II) lung cancer will relapse is vital because, despite curative-intent surgery, a significant subset of these patients harbors occult micrometastases that will lead to recurrence. Using a unique dataset curated over 10 years in Hadassah, we are working on ways to highlight patients with high risk of relapse

Identifying Graft-versus-Host Disease (GVHD) is critical because it remains a leading cause of non-relapse mortality following allogeneic hematopoietic stem cell transplantation.  Distinguishing GVHD from mimicking conditions like viral infections or drug-induced injury ensures that patients receive the correct targeted therapy while avoiding unnecessary toxicity.

Predicting the clinical course of multiple myeloma is essential for implementing a risk-stratified approach that balances treatment intensity with the patient’s specific disease biology. Identifying patients at high risk for early relapse allows for the early introduction of aggressive quadruplet regimens and maintenance therapies, potentially overcoming primary refractory disease