Quantitative imaging of cancer in the postgenomic era: Radio(geno)mics, deep learning, and habitats

Cancer, Год журнала: 2018, Номер 124(24), С. 4633 - 4649

Опубликована: Ноя. 1, 2018

Although cancer often is referred to as “a disease of the genes,” it indisputable that (epi)genetic properties individual cells are highly variable, even within same tumor. Hence, preexisting resistant clones will emerge and proliferate after therapeutic selection targets sensitive clones. Herein, authors propose quantitative image analytics, known “radiomics,” can be used quantify characterize this heterogeneity. Virtually every patient with imaged radiologically. Radiomics predicated on beliefs these images reflect underlying pathophysiologies, they converted into mineable data for improved diagnosis, prognosis, prediction, therapy monitoring. In last decade, radiomics has grown from a few laboratories worldwide enterprise. During growth, established convention, wherein large set annotated features (1‐2000 features) extracted segmented regions interest build classifier models separate patients their appropriate class (eg, indolent vs aggressive disease). An extension conventional application “deep learning,” convolutional neural networks detect most informative without human intervention. A further involves automatically segmenting subregions (“habitats”) tumors, which linked tumor pathophysiology. The goal enterprise provide informed decision support practice precision oncology.

Язык: Английский

---

A Combined Nomogram Model to Preoperatively Predict Histologic Grade in Pancreatic Neuroendocrine Tumors

Clinical Cancer Research, Год журнала: 2018, Номер 25(2), С. 584 - 594

Опубликована: Ноя. 5, 2018

The purpose of this study is to develop and validate a nomogram model combing radiomics features clinical characteristics preoperatively differentiate grade 1 2/3 tumors in patients with pancreatic neuroendocrine (pNET).Experimental Design: A total 137 who underwent contrast-enhanced CT from two hospitals were included study. the second hospital (n = 51) selected as an independent validation set. arterial phase was for feature extraction. Mann-Whitney U test least absolute shrinkage selection operator regression applied signature construction. combined developed by incorporating factors. association between Ki-67 index rate nuclear mitosis also investigated respectively. utility proposed evaluated using ROC, area under ROC curve (AUC), calibration curve, decision analysis (DCA). Kaplan-Meier (KM) used survival analysis.An eight-feature-combined constructed tumor predictor. combining stage showed best performance (training set: AUC 0.907; 0.891). DCA demonstrated usefulness nomogram. significant correlation observed mitosis, KM difference predicted groups (P 0.002).The could be useful differentiating pNETs.

Язык: Английский

---

Non-invasive measurement of PD-L1 status and prediction of immunotherapy response using deep learning of PET/CT images

Journal for ImmunoTherapy of Cancer, Год журнала: 2021, Номер 9(6), С. e002118 - e002118

Опубликована: Июнь 1, 2021

Currently, only a fraction of patients with non-small cell lung cancer (NSCLC) treated immune checkpoint inhibitors (ICIs) experience durable clinical benefit (DCB). According to NCCN guidelines, Programmed death-ligand 1 (PD-L1) expression status determined by immunohistochemistry (IHC) biopsies is the clinically approved companion biomarker trigger use ICI therapy. Based on prior work showing relationship between quantitative imaging and gene expression, we hypothesize that (radiomics) can provide an alternative surrogate for PD-L1 in decision support.

Язык: Английский

---

Identification of Non–Small Cell Lung Cancer Sensitive to Systemic Cancer Therapies Using Radiomics

Clinical Cancer Research, Год журнала: 2020, Номер 26(9), С. 2151 - 2162

Опубликована: Март 20, 2020

Abstract Purpose: Using standard-of-care CT images obtained from patients with a diagnosis of non–small cell lung cancer (NSCLC), we defined radiomics signatures predicting the sensitivity tumors to nivolumab, docetaxel, and gefitinib. Experimental Design: Data were collected prospectively analyzed retrospectively across multicenter clinical trials [nivolumab, n = 92, CheckMate017 (NCT01642004), CheckMate063 (NCT01721759); 50, CheckMate017; gefitinib, 46, (NCT00588445)]. Patients randomized training or validation cohorts using either 4:1 ratio (nivolumab: 72T:20V) 2:1 (docetaxel: 32T:18V; gefitinib: 31T:15V) ensure an adequate sample size in set. Radiomics derived quantitative analysis early tumor changes baseline first on-treatment assessment. For each patient, 1,160 features extracted largest measurable lesion. Tumors classified as treatment sensitive insensitive; reference standard was median progression-free survival (NCT01642004, NCT01721759) surgery (NCT00588445). Machine learning implemented select up four develop signature datasets applied patient classify sensitivity. Results: The predicted dataset study group AUC (95 confidence interval): 0.77 (0.55–1.00); 0.67 (0.37–0.96); 0.82 (0.53–0.97). serial radiographic measurements, magnitude exponential increase deciphering volume, invasion boundaries, spatial heterogeneity associated shorter overall survival. Conclusions: NSCLC, offering approach that could enhance decision-making continue systemic therapies forecast

Язык: Английский

---

Artificial intelligence in cancer diagnosis and therapy: Current status and future perspective

Computers in Biology and Medicine, Год журнала: 2023, Номер 165, С. 107356 - 107356

Опубликована: Авг. 14, 2023

Язык: Английский

---