Saad Nadeem’s research while affiliated with Memorial Sloan Kettering Cancer Center and other places

We introduce a new AI-ready computational pathology dataset containing restained and co-registered digitized images from eight head-and-neck squamous cell carcinoma patients. Specifically, the same tumor sections were stained with the expensive multiplex immunofluorescence (mIF) assay first and then restained with cheaper multiplex immunohistochemistry (mIHC). This is a first public dataset that demonstrates the equivalence of these two staining methods which in turn allows several use cases; due to the equivalence, our cheaper mIHC staining protocol can offset the need for expensive mIF staining/scanning which requires highly-skilled lab technicians. As opposed to subjective and error-prone immune cell annotations from individual pathologists (disagreement > 50%) to drive SOTA deep learning approaches, this dataset provides objective immune and tumor cell annotations via mIF/mIHC restaining for more reproducible and accurate characterization of tumor immune microenvironment (e.g. for immunotherapy). We demonstrate the effectiveness of this dataset in three use cases: (1) IHC quantification of CD3/CD8 tumor-infiltrating lymphocytes via style transfer, (2) virtual translation of cheap mIHC stains to more expensive mIF stains, and (3) virtual tumor/immune cellular phenotyping on standard hematoxylin images. The dataset is available at https://github.com/nadeemlab/DeepLIIF.

Aims The International Medullary Thyroid Carcinoma Grading System, introduced in 2022, mandates evaluation of the Ki67 proliferation index to assign a histological grade for medullary thyroid carcinoma. However, manual counting remains a tedious and time‐consuming task. Methods and results We aimed to evaluate the performance of three other counting techniques for the Ki67 index, eyeballing by a trained experienced investigator, a machine learning‐based deep learning algorithm (DeepLIIF) and an image analysis software with internal thresholding compared to the gold standard manual counting in a large cohort of 260 primarily resected medullary thyroid carcinoma. The Ki67 proliferation index generated by all three methods correlate near‐perfectly with the manual Ki67 index, with kappa values ranging from 0.884 to 0.979 and interclass correlation coefficients ranging from 0.969 to 0.983. Discrepant Ki67 results were only observed in cases with borderline manual Ki67 readings, ranging from 3 to 7%. Medullary thyroid carcinomas with a high Ki67 index (≥ 5%) determined using any of the four methods were associated with significantly decreased disease‐specific survival and distant metastasis‐free survival. Conclusions We herein validate a machine learning‐based deep‐learning platform and an image analysis software with internal thresholding to generate accurate automatic Ki67 proliferation indices in medullary thyroid carcinoma. Manual Ki67 count remains useful when facing a tumour with a borderline Ki67 proliferation index of 3–7%. In daily practice, validation of alternative evaluation methods for the Ki67 index in MTC is required prior to implementation.

We propose an interactive visual analytics tool, Vis-SPLIT, for partitioning a population of individuals into groups with similar gene signatures. Vis-SPLIT allows users to interactively explore a dataset and exploit visual separations to build a classification model for specific cancers. The visualization components reveal gene expression and correlation to assist specific partitioning decisions, while also providing overviews for the decision model and clustered genetic signatures. We demonstrate the effectiveness of our framework through a case study and evaluate its usability with domain experts. Our results show that Vis-SPLIT can classify patients based on their genetic signatures to effectively gain insights into RNA sequencing data, as compared to an existing classification system.

As predictive biomarkers of response to immune checkpoint inhibitors (ICIs) remain a major unmet clinical need in patients with urothelial carcinoma (UC), we sought to identify tissue-based immune biomarkers of clinical benefit to ICIs using multiplex immunofluorescence and to integrate these findings with previously identified peripheral blood biomarkers of response. Fifty-five pretreatment and 12 paired on-treatment UC specimens were identified from patients treated with nivolumab with or without ipilimumab. Whole tissue sections were stained with a 12-plex mIF panel, including CD8, PD-1/CD279, PD-L1/CD274, CD68, CD3, CD4, FoxP3, TCF1/7, Ki67, LAG-3, MHC-II/HLA-DR, and pancytokeratin+SOX10 to identify over three million cells. Immune tissue densities were compared to progression-free survival (PFS) and best overall response (BOR) by RECIST version 1.1. Correlation coefficients were calculated between tissue-based and circulating immune populations. The frequency of intratumoral CD3+ LAG-3+ cells was higher in responders compared to nonresponders (p = 0.0001). LAG-3+ cellular aggregates were associated with response, including CD3+ LAG-3+ in proximity to CD3+ (p = 0.01). Exploratory multivariate modeling showed an association between intratumoral CD3+ LAG-3+ cells and improved PFS independent of prognostic clinical factors (log HR -7.0; 95% confidence interval [CI] -12.7 to -1.4), as well as established biomarkers predictive of ICI response (log HR -5.0; 95% CI -9.8 to -0.2). Intratumoral LAG-3+ immune cell populations warrant further study as a predictive biomarker of clinical benefit to ICIs. Differences in LAG-3+ lymphocyte populations across the intratumoral and peripheral compartments may provide complementary information that could inform the future development of multimodal composite biomarkers of ICI response. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

We introduce a new AI-ready computational pathology dataset containing restained and co-registered digitized images from eight head-and-neck squamous cell carcinoma patients. Specifically, the same tumor sections were stained with the expensive multiplex immunofluorescence (mIF) assay first and then restained with cheaper multiplex immunohistochemistry (mIHC). This is a first public dataset that demonstrates the equivalence of these two staining methods which in turn allows several use cases; due to the equivalence, our cheaper mIHC staining protocol can offset the need for expensive mIF staining/scanning which requires highly-skilled lab technicians. As opposed to subjective and error-prone immune cell annotations from individual pathologists (disagreement > 50%) to drive SOTA deep learning approaches, this dataset provides objective immune and tumor cell annotations via mIF/mIHC restaining for more reproducible and accurate characterization of tumor immune microenvironment (e.g. for immunotherapy). We demonstrate the effectiveness of this dataset in three use cases: (1) IHC quantification of CD3/CD8 tumor-infiltrating lymphocytes via style transfer, (2) virtual translation of cheap mIHC stains to more expensive mIF stains, and (3) virtual tumor/immune cellular phenotyping on standard hematoxylin images. The dataset is available at \url{https://github.com/nadeemlab/DeepLIIF}.

We introduce a new AI-ready computational pathology dataset containing restained and co-registered digitized images from eight head-and-neck squamous cell carcinoma patients. Specifically, the same tumor sections were stained with the expensive multiplex immunofluorescence (mIF) assay first and then restained with cheaper multiplex immunohistochemistry (mIHC). This is a first public dataset that demonstrates the equivalence of these two staining methods which in turn allows several use cases; due to the equivalence, our cheaper mIHC staining protocol can offset the need for expensive mIF staining/scanning which requires highly-skilled lab technicians. As opposed to subjective and error-prone immune cell annotations from individual pathologists (disagreement > 50%) to drive SOTA deep learning approaches, this dataset provides objective immune and tumor cell annotations via mIF/mIHC restaining for more reproducible and accurate characterization of tumor immune microenvironment (e.g. for immunotherapy). We demonstrate the effectiveness of this dataset in three use cases: (1) IHC quantification of CD3/CD8 tumor-infiltrating lymphocytes via style transfer, (2) virtual translation of cheap mIHC stains to more expensive mIF stains, and (3) virtual tumor/immune cellular phenotyping on standard hematoxylin images. The dataset is available at \url{https://github.com/nadeemlab/DeepLIIF}.

Objective: This work aims to generate realistic anatomical deformations from static patient scans. Specifically, we present a method to generate these deformations/augmentations via deep learning driven respiratory motion simulation that provides the ground truth for validating deformable image registration (DIR) algorithms and driving more accurate deep learning based DIR. Approach: We present a novel 3D Seq2Seq deep learning respiratory motion simulator (RMSim) that learns from 4D-CT images and predicts future breathing phases given a static CT image. The predicted respiratory patterns, represented by time-varying displacement vector fields (DVFs) at different breathing phases, are modulated through auxiliary inputs of 1D breathing traces so that a larger amplitude in the trace results in more significant predicted deformation. Stacked 3D-ConvLSTMs are used to capture the spatial-temporal respiration patterns. Training loss includes a smoothness loss in the DVF and mean-squared error between the predicted and ground truth phase images. A spatial transformer deforms the static CT with the predicted DVF to generate the predicted phase image. 10-phase 4D-CTs of 140 internal patients were used to train and test RMSim. The trained RMSim was then used to augment a public DIR challenge dataset for training VoxelMorph to show the effectiveness of RMSim-generated deformation augmentation. Main results: We validated our RMSim output with both private and public benchmark datasets (healthy and cancer patients). The structure similarity index measure (SSIM) for predicted phase images was 0.92±0.04, demonstrating RMSim's potential to generate realistic respiratory motion. Moreover, the landmark registration error in a public DIR dataset was improved from 8.12±5.78mm to 6.58±6.38mm using RMSim-augmented training data. Significance: The proposed approach can be used for validating DIR algorithms as well as for patient-specific augmentations to improve deep learning DIR algorithms. The code, pretrained models, and augmented DIR validation datasets will be released at https://github.com/nadeemlab/SeqX2Y.