Yuchae Jung

Background/Objectives: How the histologically benign tier of lung preinvasive lesions—atypical adenomatous hyperplasia (AAH) and squamous dysplasia (SD)—relates genomically to its paired carcinoma is unclear. To identify early versus late events, we compared synchronous preinvasive and invasive lesions from the same patient. Methods: Whole-exome sequencing was performed on 33 FFPE samples from 11 patients (7 AAH–lung adenocarcinoma [LUAD] and 4 SD–squamous cell carcinoma [SqCC] pairs, with paired normal lung). FFPE artefacts were mitigated by paired-normal subtraction, panel-of-normals filtering, and orthogonal caller cross-validation. Cancer-panel variants were classified as cancer-only, shared, or preinvasive-only. Results: Only ∼10% of cancer-panel variants were shared between paired lesions (∼50% carcinoma-only, ∼40% preinvasive-only), indicating that benign AAH/SD do not broadly mirror the paired carcinoma. Within this small shared fraction, the early-driver pattern diverged between tracks: AAH–LUAD pairs tended to share EGFR alterations, whereas the four SD–SqCC pairs featured MET-pathway alterations without any shared EGFR events. TP53 and most other canonical drivers were carcinoma-confined (within-cohort contrast direction-consistent but non-significant); three patients lacked any canonical driver despite substantial mutational burden. Conclusions: In this pilot cohort, benign AAH and SD were genomically largely distinct from their paired carcinomas, sharing only a small set of key drivers whose identity diverged between glandular and squamous tracks. This suggests that benign-appearing AAH/SD differ from the more advanced AIS/MIA precursors not only histologically but also at the genomic level. These hypothesis-generating findings require confirmation in larger, multi-omic cohorts.

Despite remarkable advances in neural language models, a substantial gap remains in precisely interpreting the complex semantics of Electronic Medical Records (EMR). We propose Contrastive Representations Pre-Training (CRPT) to address this gap, replacing the conventional Next Sentence Prediction task’s cross-entropy loss with contrastive loss and incorporating whole-word masking to capture multi-token domain-specific terms better. We also introduce a carefully designed negative sampling strategy that balances intra-document and cross-document sentences, enhancing the model’s discriminative power. Implemented atop a BERT-based architecture and evaluated on the Biomedical Language Understanding Evaluation (BLUE) benchmark, our Discharge Summary CRPT model achieves significant performance gains, including a natural language inference precision of 0.825 and a sentence similarity score of 0.775. We further extend our approach through Bio+Discharge Summary CRPT, combining biomedical and clinical corpora to boost downstream performance across tasks. Our framework demonstrates robust interpretive capacity in clinical texts by emphasizing sentence-level semantics and domain-aware masking. These findings underscore CRPT’s potential for advancing semantic accuracy in healthcare applications and open new avenues for integrating larger negative sample sets, domain-specific masking techniques, and multi-task learning paradigms.

Diffuse gliomas are the most common primary brain tumors and they vary considerably in their morphology, location, genetic alterations, and response to therapy. In 2016, the World Health Organization (WHO) provided new guidelines for making an integrated diagnosis that incorporates both morphologic and molecular features to diffuse gliomas. In this study, we demonstrate how deep learning approaches can be used for an automatic classification of glioma subtypes and grading using whole-slide images that were obtained from routine clinical practice. A deep transfer learning method using the ResNet50V2 model was trained to classify subtypes and grades of diffuse gliomas according to the WHO’s new 2016 classification. The balanced accuracy of the diffuse glioma subtype classification model with majority voting was 0.8727. These results highlight an emerging role of deep learning in the future practice of pathologic diagnosis.