Sajid Javed presents Cancerous Footprints in Histopathological landscape
On 2025-07-24 13:00:00 at G205, Karlovo náměstí 13, Praha 2
Computational pathology plays an important role in the clinical practices for
better cancer diagnoses and prognostications. Expert pathologists harness
computational pathology tools to diagnose and grade cancers and even predict
the
cancer biomarkers using gigapixel histopathological whole slide images. In the
literature, computational pathology has been evolved in terms of classifying
whole slide images, segmenting whole slide images, and discovery of the new
cancer biomarkers. Thanks to the computational pathology community for
introducing new insights and large-scale datasets. Multi-modal large language
models in computational pathology have opened new directions specifically these
models assist expert pathologists to reason the histopathology images.
In this talk, I will discuss two computational pathology foundational models
e.g., CPLIP and MR-PLIP, recently developed for histology image analysis.
CPLIP
is pre-trained in a completely unsupervised learning manners for performing
zero-shot tasks including classification and segmentation while MR-PLIP has
been
pre-trained by utilizing multi-resolution hierarchy of the whole slide images.
Both models showed strong generalization capabilities compared to the
state-of-the-art pathology models.
better cancer diagnoses and prognostications. Expert pathologists harness
computational pathology tools to diagnose and grade cancers and even predict
the
cancer biomarkers using gigapixel histopathological whole slide images. In the
literature, computational pathology has been evolved in terms of classifying
whole slide images, segmenting whole slide images, and discovery of the new
cancer biomarkers. Thanks to the computational pathology community for
introducing new insights and large-scale datasets. Multi-modal large language
models in computational pathology have opened new directions specifically these
models assist expert pathologists to reason the histopathology images.
In this talk, I will discuss two computational pathology foundational models
e.g., CPLIP and MR-PLIP, recently developed for histology image analysis.
CPLIP
is pre-trained in a completely unsupervised learning manners for performing
zero-shot tasks including classification and segmentation while MR-PLIP has
been
pre-trained by utilizing multi-resolution hierarchy of the whole slide images.
Both models showed strong generalization capabilities compared to the
state-of-the-art pathology models.