Simulated deep CT characterization of liver metastases with high-resolution filtered back projection reconstruction.

Early diagnosis and accurate prognosis of colorectal cancer is critical for determining optimal treatment plans and maximizing patient outcomes, especially as the disease progresses into liver metastases. Computed tomography (CT) is a frontline tool for this task; however, the preservation of predictive radiomic features is highly dependent on the scanning protocol and reconstruction algorithm. We hypothesized that image reconstruction with a high-frequency kernel could result in a better characterization of liver metastases features via deep neural networks. This kernel produces images that appear noisier but preserve more sinogram information. A simulation pipeline was developed to study the effects of imaging parameters on the ability to characterize the features of liver metastases. This pipeline utilizes a fractal approach to generate a diverse population of shapes representing virtual metastases, and then it superimposes them on a realistic CT liver region to perform a virtual CT scan using CatSim. Datasets of 10,000 liver metastases were generated, scanned, and reconstructed using either standard or high-frequency kernels. These data were used to train and validate deep neural networks to recover crafted metastases characteristics, such as internal heterogeneity, edge sharpness, and edge fractal dimension. In the absence of noise, models scored, on average, 12.2% ( α = 0.012 ) and 7.5% ( α = 0.049 ) lower squared error for characterizing edge sharpness and fractal dimension, respectively, when using high-frequency reconstructions compared to standard. However, the differences in performance were statistically insignificant when a typical level of CT noise was simulated in the clinical scan. Our results suggest that high-frequency reconstruction kernels can better preserve information for downstream artificial intelligence-based radiomic characterization, provided that noise is limited. Future work should investigate the information-preserving kernels in datasets with clinical labels.

Multiplexed Imaging Mass Cytometry Analysis Characterizes the Vascular Niche in Pancreatic Cancer.

Oncogenesis and progression of pancreatic ductal adenocarcinoma (PDAC) is driven by complex interactions between the neoplastic component and the tumor microenvironment (TME), which includes immune, stromal, and parenchymal cells. In particular, most PDACs are characterized by a hypovascular and hypoxic environment that alters tumor cell behavior and limits the efficacy of chemotherapy and immunotherapy. Characterization of the spatial features of the vascular niche could advance our understanding of inter- and intra-tumoral heterogeneity in PDAC. Here, we investigated the vascular microenvironment of PDAC by applying imaging mass cytometry using a 26-antibody panel on 35 regions of interest (ROIs) across 9 patients, capturing over 140,000 single cells. The approach distinguished major cell types, including multiple populations of lymphoid and myeloid cells, endocrine cells, ductal cells, stromal cells, and endothelial cells. Evaluation of cellular neighborhoods identified 10 distinct spatial domains, including multiple immune and tumor-enriched environments as well as the vascular niche. Focused analysis revealed differential interactions between immune populations and the vasculature and identified distinct spatial domains wherein tumor cell proliferation occurs. Importantly, the vascular niche was closely associated with a population of CD44-expressing macrophages enriched for a pro-angiogenic gene signature. Together, this study provides insights into the spatial heterogeneity of PDAC and suggests a role for CD44-expressing macrophages in shaping the vascular niche.

Tumor Size Is Not Everything: Advancing Radiomics as a Precision Medicine Biomarker in Oncology Drug Development and Clinical Care. A Report of a Multidisciplinary Workshop Coordinated by the RECIST Working Group.

Radiomics, the science of extracting quantifiable data from routine medical images, is a powerful tool that has many potential applications in oncology. The Response Evaluation Criteria in Solid Tumors Working Group (RWG) held a workshop in May 2022, which brought together various stakeholders to discuss the potential role of radiomics in oncology drug development and clinical trials, particularly with respect to response assessment. This article summarizes the results of that workshop, reviewing radiomics for the practicing oncologist and highlighting the work that needs to be done to move forward the incorporation of radiomics into clinical trials.

Use of Response Permutation to Measure an Imaging Dataset's Susceptibility to Overfitting by Selected Standard Analysis Pipelines.

RATIONALE AND OBJECTIVES: This study demonstrates a method for quantifying the impact of overfitting on the receiving operator characteristic curve (AUC) when using standard analysis pipelines to develop imaging biomarkers. We illustrate the approach using two publicly available repositories of radiology and pathology images for breast cancer diagnosis. MATERIALS AND METHODS: For each dataset, we permuted the outcome (cancer diagnosis) values to eliminate any true association between imaging features and outcome. Seven types of classification models (logistic regression, linear discriminant analysis, Naïve Bayes, linear support vector machines, nonlinear support vector machine, random forest, and multi-layer perceptron) were fitted to each scrambled dataset and evaluated by each of four techniques (all data, hold-out, 10-fold cross-validation, and bootstrapping). After repeating this process for a total of 50 outcome permutations, we averaged the resulting AUCs. Any increase over a null AUC of 0.5 can be attributed to overfitting. RESULTS: Applying this approach and varying sample size and the number of imaging features, we found that failing to control for overfitting could result in near-perfect prediction (AUC near 1.0). Cross-validation offered greater protection against overfitting than the other evaluation techniques, and for most classification algorithms a sample size of at least 200 was required to assess as few as 10 features with less than 0.05 AUC inflation attributable to overfitting. CONCLUSION: This approach could be applied to any curated dataset to suggest the number of features and analysis approaches to limit overfitting.

Mechanisms of Exercise Intolerance Across the Breast Cancer Continuum: A Pooled Analysis of Individual Patient Data.

PURPOSE: The purpose of this study is to evaluate the prevalence of abnormal cardiopulmonary responses to exercise and pathophysiological mechanism(s) underpinning exercise intolerance across the continuum of breast cancer (BC) care from diagnosis to metastatic disease. METHODS: Individual participant data from four randomized trials spanning the BC continuum ([1] prechemotherapy [n = 146], [2] immediately postchemotherapy [n = 48], [3] survivorship [n = 138], and [4] metastatic [n = 47]) were pooled and compared with women at high-risk of BC (BC risk; n = 64). Identical treadmill-based peak cardiopulmonary exercise testing protocols evaluated exercise intolerance (peak oxygen consumption; V̇O2peak) and other resting, submaximal, and peak cardiopulmonary responses. The prevalence of 12 abnormal exercise responses was evaluated. Graphical plots of exercise responses were used to identify oxygen delivery and/or uptake mechanisms contributing to exercise intolerance. Unsupervised, hierarchical cluster analysis was conducted to explore exercise response phenogroups. RESULTS: Mean V̇O2peak was 2.78 ml O2.kg-1·min-1 (95% confidence interval [CI], -3.94, -1.62 mL O2.kg-1·min-1; P < 0.001) lower in the pooled BC cohort (52 ± 11 yr) than BC risk (55 ± 10 yr). Compared with BC risk, the pooled BC cohort had a 2.5-fold increased risk of any abnormal cardiopulmonary response (odds ratio, 2.5; 95% confidence interval, 1.2, 5.3; P = 0.014). Distinct exercise responses in BC reflected impaired oxygen delivery and uptake relative to control, although considerable inter-individual heterogeneity within cohorts was observed. In unsupervised, hierarchical cluster analysis, six phenogroups were identified with marked differences in cardiopulmonary response patterns and unique clinical characteristics. CONCLUSIONS: Abnormal cardiopulmonary response to exercise is common in BC and is related to impairments in oxygen delivery and uptake. The identification of exercise response phenogroups could help improve cardiovascular risk stratification and guide investigation of targeted exercise interventions.

Preoperative CT and survival data for patients undergoing resection of colorectal liver metastases.

The liver is a common site for the development of metastases in colorectal cancer. Treatment selection for patients with colorectal liver metastases (CRLM) is difficult; although hepatic resection will cure a minority of CRLM patients, recurrence is common. Reliable preoperative prediction of recurrence could therefore be a valuable tool for physicians in selecting the best candidates for hepatic resection in the treatment of CRLM. It has been hypothesized that evidence for recurrence could be found via quantitative image analysis on preoperative CT imaging of the future liver remnant before resection. To investigate this hypothesis, we have collected preoperative hepatic CT scans, clinicopathologic data, and recurrence/survival data, from a large, single-institution series of patients (n = 197) who underwent hepatic resection of CRLM. For each patient, we also created segmentations of the liver, vessels, tumors, and future liver remnant. The largest of its kind, this dataset is a resource that may aid in the development of quantitative imaging biomarkers and machine learning models for the prediction of post-resection hepatic recurrence of CRLM.

Medical image synthesis via conditional GANs: Application to segmenting brain tumours.

Accurate brain tumour segmentation is critical for tasks such as surgical planning, diagnosis, and analysis, with magnetic resonance imaging (MRI) being the preferred modality due to its excellent visualisation of brain tissues. However, the wide intensity range of voxel values in MR scans often results in significant overlap between the density distributions of different tumour tissues, leading to reduced contrast and segmentation accuracy. This paper introduces a novel framework based on conditional generative adversarial networks (cGANs) aimed at enhancing the contrast of tumour subregions for both voxel-wise and region-wise segmentation approaches. We present two models: Enhancement and Segmentation GAN (ESGAN), which combines classifier loss with adversarial loss to predict central labels of input patches, and Enhancement GAN (EnhGAN), which generates high-contrast synthetic images with reduced inter-class overlap. These synthetic images are then fused with corresponding modalities to emphasise meaningful tissues while suppressing weaker ones. We also introduce a novel generator that adaptively calibrates voxel values within input patches, leveraging fully convolutional networks. Both models employ a multi-scale Markovian network as a GAN discriminator to capture local patch statistics and estimate the distribution of MR images in complex contexts. Experimental results on publicly available MR brain tumour datasets demonstrate the competitive accuracy of our models compared to current brain tumour segmentation techniques.

Machine Learning Detection and Characterization of Splenic Injuries on Abdominal Computed Tomography.

BACKGROUND: Multi-detector contrast-enhanced abdominal computed tomography (CT) allows for the accurate detection and classification of traumatic splenic injuries, leading to improved patient management. Their effective use requires rapid study interpretation, which can be a challenge on busy emergency radiology services. A machine learning system has the potential to automate the process, potentially leading to a faster clinical response. This study aimed to create such a system. METHOD: Using the American Association for the Surgery of Trauma (AAST), spleen injuries were classified into 3 classes: normal, low-grade (AAST grade I-III) injuries, and high-grade (AAST grade IV and V) injuries. Employing a 2-stage machine learning strategy, spleens were initially segmented from input CT images and subsequently underwent classification via a 3D dense convolutional neural network (DenseNet). RESULTS: This single-centre retrospective study involved trauma protocol CT scans performed between January 1, 2005, and July 31, 2021, totaling 608 scans with splenic injuries and 608 without. Five board-certified fellowship-trained abdominal radiologists utilizing the AAST injury scoring scale established ground truth labels. The model achieved AUC values of 0.84, 0.69, and 0.90 for normal, low-grade injuries, and high-grade splenic injuries, respectively. CONCLUSIONS: Our findings demonstrate the feasibility of automating spleen injury detection using our method with potential applications in improving patient care through radiologist worklist prioritization and injury stratification. Future endeavours should concentrate on further enhancing and optimizing our approach and testing its use in a real-world clinical environment.

Identifying and shifting educators' failure pedagogical mindsets through reflective practices.

BACKGROUND: In this paper, we add to the scant literature base on learning from failures with a particular focus on understanding educators' shifting mindset in making-centred learning environments. AIMS: The aim of Study 1 was to explore educators' beliefs about failure for learning and instructional practices within their local making-centred learning environments. The aim of Study 2 was to examine how participation in a video-based professional development cycle regarding failure moments in making-centred learning environments might have shifted museum educators' failure pedagogical mindsets. SAMPLE: In Study 1, the sample included 15 educators at either a middle school or museum. In Study 2, the sample included 39 educators across six museums. METHODS: In Study 1, educators engaged in a semi-structured interview that lasted between 45 and 75 min. In Study 2, the six museums video recorded professional development sessions. RESULTS: Results from Study 1 highlighted educators' failure pedagogical mindsets as either underdeveloped or rigid and absent of relational thinking between self- and youth-failures. One key result from Study 2 was a shift from an abstract sense of failure as youth-focused to a practical sense of failure as educator-focused and/or relational (i.e., youth educator-focused failure moments). CONCLUSIONS: Based on the results from Study 1 and Study 2, our research suggests that exploring an educator's relationship with failure is important and witnessing and reflecting upon their own failure pedagogical mindset in action may facilitate a shift towards a more complex and interconnected space for growth and development of both educators and youth.

The RSNA Cervical Spine Fracture CT Dataset.

This dataset is composed of cervical spine CT images with annotations related to fractures; it is available at https://www.kaggle.com/competitions/rsna-2022-cervical-spine-fracture-detection/.

Applying Natural Language Processing to Single-Report Prediction of Metastatic Disease Response Using the OR-RADS Lexicon.

Generating Real World Evidence (RWE) on disease responses from radiological reports is important for understanding cancer treatment effectiveness and developing personalized treatment. A lack of standardization in reporting among radiologists impacts the feasibility of large-scale interpretation of disease response. This study examines the utility of applying natural language processing (NLP) to the large-scale interpretation of disease responses using a standardized oncologic response lexicon (OR-RADS) to facilitate RWE collection. Radiologists annotated 3503 retrospectively collected clinical impressions from radiological reports across several cancer types with one of seven OR-RADS categories. A Bidirectional Encoder Representations from Transformers (BERT) model was trained on this dataset with an 80-20% train/test split to perform multiclass and single-class classification tasks using the OR-RADS. Radiologists also performed the classification to compare human and model performance. The model achieved accuracies from 95 to 99% across all classification tasks, performing better in single-class tasks compared to the multiclass task and producing minimal misclassifications, which pertained mostly to overpredicting the equivocal and mixed OR-RADS labels. Human accuracy ranged from 74 to 93% across all classification tasks, performing better on single-class tasks. This study demonstrates the feasibility of the BERT NLP model in predicting disease response in cancer patients, exceeding human performance, and encourages the use of the standardized OR-RADS lexicon to improve large-scale prediction accuracy.

The Association between COVID-19 and Changes in Opioid Prescribing Patterns and Opioid-Related Overdoses: A Retrospective Cohort Study.

Background: Recent data suggest that restrictions related to COVID-19 resulted in changes in the prescribing patterns of opioids. Aims: We sought to analyze Ontario health data for changes in frequencies among new and continuing users for the following opioid prescription characteristics: the type of opioid, the average daily dose, and the prescriber's specialty. Methods: Utilizing data on the Ontario Health Data Platform, we defined two 149-day windows as "before" and "after" based on the initial COVID-19 provincial lockdown. A total of 882,268 individuals met our inclusion criteria and were classified as either "new" or "continuing" users. Chi-square tests and Fisher's exact tests were applied for each level of our primary outcomes to determine whether there were significant changes in prescription proportions before and after the lockdown. Results: A decline of 28% was observed for the number of new users after the lockdown. Statistically significant changes were observed for new users across almost all opioid prescription characteristics between the before and after windows. The proportion of new users who received at least one dispensing event from a pharmacist increased by 26.32%, whereas continuing users increased by 378.61%. There were no statistically significant shifts in opioid prescriptions among individuals with a reported toxicity event during the study period. Conclusions: In terms of opioid prescribing patterns, new users experienced greater change following the onset of the pandemic lockdown than continuing users. Our findings potentially showcase the unintended impacts that COVID-19-related restrictions had on non-COVID-19-related health services, which can inform future policy decisions.

Contexte: Des données récentes indiquent que les restrictions liées à la COVID-19 ont entrainé des changements dans la prescription des opioïdes.Objectifs: Nous avons cherché à analyser les données sur la santé de l'Ontario pour déceler les changements de fréquence chez les nouveaux utilisateurs et les utilisateurs prévalents pour les caractéristiques de prescription d'opioïdes suivantes : le type d'opioïde, la dose quotidienne moyenne, et la spécialité du prescripteur.Méthodes: En utilisant les données de la plateforme de données sur la santé de l'Ontario, nous avons défini deux fenêtres de 149 jours comme suit : « avant ¼ et « après ¼ le confinement provincial initial de la COVID-19. Un total de 882 268 personnes ont répondu à nos critères d'inclusion et ont été classées comme « nouveaux utilisateurs ¼ ou « utilisateurs prévalents ¼. Des tests de chi-carré et des tests exacts de Fisher ont été appliqués pour chaque niveau de nos résultats primaires afin de déterminer s'il y avait eu des changements importants dans les proportions prescrites avant et après le confinement.Résultats: Une baisse de 28 % a été observée pour le nombre de nouveaux utilisateurs après le confinement.Des changements statistiquement significatifs ont été observés pour les nouveaux utilisateurs pour presque toutes les caractéristiques de prescriptions d'opioïdes entre les fenêtres avant et après. La proportion de nouveaux utilisateurs ayant eu au moins une prescription remplie par un pharmacien a augmenté de 26,32 %, tandis que le nombre d'utilisateurs prévalents a augmenté de 378,61 %. Il n'y a pas eu de changements statistiquement significatifs dans les prescriptions d'opioïdes parmi les personnes ayant déclaré un évènement de ⁠toxicité au cours de la période d'étude.Conclusions: En matière de modèles de prescription d'opioïdes, les nouveaux utilisateurs ont connu un changement plus important après le début du confinement de la pandémie que les utilisateurs prévalents. Nos résultats démontrent possiblement les répercussions inattendues des restrictions liées à la COVID-19 sur les services de santé non liés à la COVID-19, ce qui pourrait éclairer les décisions politiques futures.

Computerized Diagnosis of Liver Tumors From CT Scans Using a Deep Neural Network Approach.

The liver is a frequent site of benign and malignant, primary and metastatic tumors. Hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) are the most common primary liver cancers, and colorectal liver metastasis (CRLM) is the most common secondary liver cancer. Although the imaging characteristic of these tumors is central to optimal clinical management, it relies on imaging features that are often non-specific, overlap, and are subject to inter-observer variability. Thus, in this study, we aimed to categorize liver tumors automatically from CT scans using a deep learning approach that objectively extracts discriminating features not visible to the naked eye. Specifically, we used a modified Inception v3 network-based classification model to classify HCC, ICC, CRLM, and benign tumors from pretreatment portal venous phase computed tomography (CT) scans. Using a multi-institutional dataset of 814 patients, this method achieved an overall accuracy rate of 96%, with sensitivity rates of 96%, 94%, 99%, and 86% for HCC, ICC, CRLM, and benign tumors, respectively, using an independent dataset. These results demonstrate the feasibility of the proposed computer-assisted system as a novel non-invasive diagnostic tool to classify the most common liver tumors objectively.

Natural Language Processing of Computed Tomography Reports to Label Metastatic Phenotypes With Prognostic Significance in Patients With Colorectal Cancer.

PURPOSE: Natural language processing (NLP) applied to radiology reports can help identify clinically relevant M1 subcategories of patients with colorectal cancer (CRC). The primary purpose was to compare the overall survival (OS) of CRC according to American Joint Committee on Cancer TNM staging and explore an alternative classification. The secondary objective was to estimate the frequency of metastasis for each organ. METHODS: Retrospective study of CRC who underwent computed tomography (CT) chest, abdomen, and pelvis between July 1, 2009, and March 26, 2019, at a tertiary cancer center, previously labeled for the presence or absence of metastasis by an NLP prediction model. Patients were classified in M0, M1a, M1b, and M1c (American Joint Committee on Cancer), or an alternative classification on the basis of the metastasis organ number: M1, single; M2, two; M3, three or more organs. Cox regression models were used to estimate hazard ratios; Kaplan-Meier curves were used to visualize survival curves using the two M1 subclassifications. RESULTS: Nine thousand nine hundred twenty-eight patients with a total of 48,408 CT chest, abdomen, and pelvis reports were included. On the basis of NLP prediction, the median OS of M1a, M1b, and M1c was 4.47, 1.72, and 1.52 years, respectively. The median OS of M1, M2, and M3 was 4.24, 2.05, and 1.04 years, respectively. Metastases occurred most often in liver (35.8%), abdominopelvic lymph nodes (32.9%), lungs (29.3%), peritoneum (22.0%), thoracic nodes (19.9%), bones (9.2%), and pelvic organs (7.5%). Spleen and adrenal metastases occurred in < 5%. CONCLUSION: NLP applied to a large radiology report database can identify clinically relevant metastatic phenotypes and be used to investigate new M1 substaging for CRC. Patients with three or more metastatic disease organs have the worst prognosis, with an OS of 1 year.

Radiomics artificial intelligence modelling for prediction of local control for colorectal liver metastases treated with radiotherapy.

Background and Purpose: Prognostic assessment of local therapies for colorectal liver metastases (CLM) is essential for guiding management in radiation oncology. Computed tomography (CT) contains liver texture information which may be predictive of metastatic environments. To investigate the feasibility of analyzing CT texture, we sought to build an automated model to predict progression-free survival using CT radiomics and artificial intelligence (AI). Materials and Methods: Liver CT scans and outcomes for N = 97 CLM patients treated with radiotherapy were retrospectively obtained. A survival model was built by extracting 108 radiomic features from liver and tumor CT volumes for a random survival forest (RSF) to predict local progression. Accuracies were measured by concordance indices (C-index) and integrated Brier scores (IBS) with 4-fold cross-validation. This was repeated with different liver segmentations and radiotherapy clinical variables as inputs to the RSF. Predictive features were identified by perturbation importances. Results: The AI radiomics model achieved a C-index of 0.68 (CI: 0.62-0.74) and IBS below 0.25 and the most predictive radiomic feature was gray tone difference matrix strength (importance: 1.90 CI: 0.93-2.86) and most predictive treatment feature was maximum dose (importance: 3.83, CI: 1.05-6.62). The clinical data only model achieved a similar C-index of 0.62 (CI: 0.56-0.69), suggesting that predictive signals exist in radiomics and clinical data. Conclusions: The AI model achieved good prediction accuracy for progression-free survival of CLM, providing support that radiomics or clinical data combined with machine learning may aid prognostic assessment and management.

Radiomic Analysis: Study Design, Statistical Analysis, and Other Bias Mitigation Strategies.

Rapid advances in automated methods for extracting large numbers of quantitative features from medical images have led to tremendous growth of publications reporting on radiomic analyses. Translation of these research studies into clinical practice can be hindered by biases introduced during the design, analysis, or reporting of the studies. Herein, the authors review biases, sources of variability, and pitfalls that frequently arise in radiomic research, with an emphasis on study design and statistical analysis considerations. Drawing on existing work in the statistical, radiologic, and machine learning literature, approaches for avoiding these pitfalls are described.

Recurrence After Resection of Pancreatic Cancer: Can Radiomics Predict Patients at Greatest Risk of Liver Metastasis?

BACKGROUND: Liver metastasis (LM) after pancreatic ductal adenocarcinoma (PDAC) resection is common but difficult to predict and has grave prognosis. We combined preoperative clinicopathological variables and quantitative analysis of computed tomography (CT) imaging to predict early LM. METHODS: We retrospectively evaluated patients with PDAC submitted to resection between 2005 and 2014 and identified clinicopathological variables associated with early LM. We performed liver radiomic analysis on preoperative contrast-enhanced CT scans and developed a logistic regression classifier to predict early LM (< 6 months). RESULTS: In 688 resected PDAC patients, there were 516 recurrences (75%). The cumulative incidence of LM at 5 years was 41%, and patients who developed LM first (n = 194) had the lowest 1-year overall survival (OS) (34%), compared with 322 patients who developed extrahepatic recurrence first (61%). Independent predictors of time to LM included poor tumor differentiation (hazard ratio (HR) = 2.30; P < 0.001), large tumor size (HR = 1.17 per 2-cm increase; P = 0.048), lymphovascular invasion (HR = 1.50; P = 0.015), and liver Fibrosis-4 score (HR = 0.89 per 1-unit increase; P = 0.029) on multivariate analysis. A model using radiomic variables that reflect hepatic parenchymal heterogeneity identified patients at risk for early LM with an area under the receiver operating characteristic curve (AUC) of 0.71; the performance of the model was improved by incorporating preoperative clinicopathological variables (tumor size and differentiation status; AUC = 0.74, negative predictive value (NPV) = 0.86). CONCLUSIONS: We confirm the adverse survival impact of early LM after resection of PDAC. We further show that a model using radiomic data from preoperative imaging combined with tumor-related variables has great potential for identifying patients at high risk for LM and may help guide treatment selection.

Developing a Cancer Digital Twin: Supervised Metastases Detection From Consecutive Structured Radiology Reports.

The development of digital cancer twins relies on the capture of high-resolution representations of individual cancer patients throughout the course of their treatment. Our research aims to improve the detection of metastatic disease over time from structured radiology reports by exposing prediction models to historical information. We demonstrate that Natural language processing (NLP) can generate better weak labels for semi-supervised classification of computed tomography (CT) reports when it is exposed to consecutive reports through a patient's treatment history. Around 714,454 structured radiology reports from Memorial Sloan Kettering Cancer Center adhering to a standardized departmental structured template were used for model development with a subset of the reports included for validation. To develop the models, a subset of the reports was curated for ground-truth: 7,732 total reports in the lung metastases dataset from 867 individual patients; 2,777 reports in the liver metastases dataset from 315 patients; and 4,107 reports in the adrenal metastases dataset from 404 patients. We use NLP to extract and encode important features from the structured text reports, which are then used to develop, train, and validate models. Three models-a simple convolutional neural network (CNN), a CNN augmented with an attention layer, and a recurrent neural network (RNN)-were developed to classify the type of metastatic disease and validated against the ground truth labels. The models use features from consecutive structured text radiology reports of a patient to predict the presence of metastatic disease in the reports. A single-report model, previously developed to analyze one report instead of multiple past reports, is included and the results from all four models are compared based on accuracy, precision, recall, and F1-score. The best model is used to label all 714,454 reports to generate metastases maps. Our results suggest that NLP models can extract cancer progression patterns from multiple consecutive reports and predict the presence of metastatic disease in multiple organs with higher performance when compared with a single-report-based prediction. It demonstrates a promising automated approach to label large numbers of radiology reports without involving human experts in a time- and cost-effective manner and enables tracking of cancer progression over time.