Sohrab P. Shah | Searcharxiv

arXiv2024

Improving ovarian cancer segmentation accuracy with transformers through AI-guided labeling

Transformer models have demonstrated the capability to produce highly accurate segmentation of organs and tumors. However, model training requires high-quality curated datasets to ensure robust generalization to unseen datasets. Hence, we developed an artificial intelligence (AI) guided approach to assist with radiologist tumor delineation of partially segmented computed tomography datasets containing primary (adnexa) tumors and metastatic (omental) implants. AI guidance was implemented by training a 2D multiple resolution residual network trained with a dataset of 245 contrast-enhanced CTs with partially segmented examples. The same dataset curated through AI guidance was then used to refine two pretrained transformer models called SMIT and Swin UNETR. The models were independently tested on 71 publicly available multi-institutional 3D CT datasets. Segmentation accuracy was computed using the Dice similarity coefficient metric (DSC), average symmetric surface distance (ASSD), and the relative volume difference (RVD) metrics. Radiomic features reproducibility was assessed using the concordance correlation coefficient (CCC). Training with AI-guided segmentations significantly improved the accuracy of both SMIT (p = 6.2e-5) and Swin UNETR (p = 2e-4) models compared with using a partially delineated training dataset. Furthermore, SMIT-generated segmentations resulted in more reproducible features compared to Swin UNETR under multiple feature categories. Our results show that AI-guided data curation provides a more efficient approach to train AI models and that AI-generated segmentations can provide reproducible radiomics features.

eess.IV↗

arXiv2015

Joint Inference of Genome Structure and Content in Heterogeneous Tumour Samples

For a genomically unstable cancer, a single tumour biopsy will often contain a mixture of competing tumour clones. These tumour clones frequently differ with respect to their genomic content (copy number of each gene) and structure (order of genes on each chromosome). Modern bulk genome sequencing mixes the signals of tumour clones and contaminating normal cells, complicating inference of genomic content and structure. We propose a method to unmix tumour and contaminating normal signals and jointly predict genomic structure and content of each tumour clone. We use genome graphs to represent tumour clones, and model the likelihood of the observed reads given clones and mixing proportions. Our use of haplotype blocks allows us to accurately measure allele specific read counts, and infer allele specific copy number for each clone. The proposed method is a heuristic local search based on applying incremental, locally optimal modifications of the genome graphs. Using simulated data, we show that our method predicts copy counts and gene adjacencies with reasonable accuracy.

q-bio.GN↗