Search

Identifying the presence of intravenous contrast material on CT scans is an important component of data curation for medical imaging-based artificial intelligence model development and deployment. Use of intravenous contrast material is often poorly documented in imaging metadata, necessitating impractical manual annotation by clinician experts. Authors developed a convolutional neural network (CNN)-based deep learning platform to identify intravenous contrast enhancement on CT scans. For model development and validation, authors used six independent datasets of head and neck (HN) and chest CT scans, totaling 133 480 axial two-dimensional sections from 1979 scans, which were manually annotated by clinical experts. Five CNN models were trained first on HN scans for contrast enhancement detection. Model performances were evaluated at the patient level on a holdout set and external test set. Models were then fine-tuned on chest CT data and externally validated. This study found that Digital Imaging and Communications in Medicine metadata tags for intravenous contrast material were missing or erroneous for 1496 scans (75.6%). An EfficientNetB4-based model showed the best performance, with areas under the curve (AUCs) of 0.996 and 1.0 in HN holdout ( n = 216) and external ( n = 595) sets, respectively, and AUCs of 1.0 and 0.980 in the chest holdout ( n = 53) and external ( n = 402) sets, respectively. This automated, scan-to-prediction platform is highly accurate at CT contrast enhancement detection and may be helpful for artificial intelligence model development and clinical application. Keywords: CT, Head and Neck, Supervised Learning, Transfer Learning, Convolutional Neural Network (CNN), Machine Learning Algorithms, Contrast Material Supplemental material is available for this article. © RSNA, 2022., Competing Interests: Disclosures of conflicts of interest: Z.Y. No relevant relationships. J.M.Q. No relevant relationships. A.H. Consultant for Altis Labs; shareholder in Altis Labs. R.Z. No relevant relationships. D.P. No relevant relationships. J.L. No relevant relationships. Z.Z. No relevant relationships. R.H.M. Contract/grant from ViewRay; consulting for ViewRay and AstraZeneca; payment for expert testimony from U.S. District Attorney's Office of New York. H.J.W.L.A. No relevant relationships. B.H.K. RSNA Research Scholar Award NIH K08DE030216., (© 2022 by the Radiological Society of North America, Inc.)

Purpose: Radiation therapy (RT)-associated lymphopenia may adversely affect treatment outcomes, particularly in the era of immunotherapy. We sought to determine dosimetric factors correlated with lymphopenia after palliative RT in a cohort of patients with advanced cancer treated with anti-PD-1 immune checkpoint inhibitors., Methods and Materials: We included patients with metastatic lung cancer, melanoma, or renal cell carcinoma who were treated with either pembrolizumab or nivolumab and received palliative RT to an extracranial site. Baseline and nadir absolute lymphocyte counts (ALCs) within 6 weeks of RT were recorded. Dosimetric factors were extracted from the corresponding dose-volume histograms and also used to model the dose to circulating lymphocytes via a whole-body blood flow model that simulates the spatiotemporal distribution of blood particles in major organs during RT., Results: We analyzed 55 patients who underwent 80 total courses of palliative RT; most (94%) were treated with 3-dimensional conformal RT. Doses to the whole body, bone, and large blood vessels (LBVs) were negatively correlated with the ALC nadir, with the strongest correlations seen at V15 (r s , -0.38, -0.43, and -0.37, and P  = .0004, .0001, and .0008, respectively). Doses to other organs were not significantly correlated with the ALC nadir. The modeled dose to circulating lymphocytes was also negatively correlated with the ALC nadir and percent ALC change (for D2%, r s , -0.31 and -0.44, and P  = .005 and .0001, respectively). Grade ≥3 lymphopenia was associated with LBV V15 (odds ratio [OR], 1.16; 95% CI, 1.07-1.26; P < .001), bone V15 (OR, 1.04; 95% CI, 1.01-1.08; P  = .03), body V15 (OR, 1.003; 95% CI, 1.001-1.006; P  = .008), and modeled lymphocyte dose (OR, 1.45; 95% CI, 1.16-1.82; P < .001)., Conclusions: The RT dose to the whole body, bone, and LBVs and the modeled dose to circulating lymphocytes were correlated with lymphopenia in patients treated with palliative RT and anti-PD-1 immune checkpoint inhibitors. These findings may inform future radiation planning in this setting., (© 2021 The Authors.)

Objective: Concurrent use of anti-PD-1 therapies with stereotactic radiosurgery (SRS) have been shown to be beneficial for survival and local lesional control in melanoma patients with brain metastases. It is not known, however, if immunotherapy (IT) confers the same outcome advantage in lung cancer patients with brain metastases treated with SRS., Methods: The authors retrospectively reviewed 85 non-small cell lung cancer (NSCLC) patients with brain metastases who were treated with SRS between January 2006 and December 2016. Thirty-nine PD-L1 antibody-positive patients received anti-PD-1 therapy with SRS (IT group) and 46 patients received chemotherapy (CT) with SRS (CT group). Results were obtained using chi-square, Kaplan-Meier, and Mann-Whitney U tests and Cox regression analyses., Results: Median survival following first radiosurgical treatment in the whole study group was 11.6 months (95% CI 8-15.5 months). Median survival times in the IT group and CT group were 10 months (95% CI 8.3-13.2 months) and 11.6 months (95% CI 7.7-15.6 months), respectively (p = 0.23). A Karnofsky Performance Status (KPS) score < 80 (p = 0.001) and lung-specific molecular marker Graded Prognostic Assessment (lungmol GPA) score < 1.5 (p = 0.02) were found to be predictive of worse survival.Maximal percent lesional shrinkage and time to maximal shrinkage were not significantly different between the CT and IT groups. Of the lesions for which a complete response occurred, 94.8% had pre-SRS volumes < 500 mm3. The amount of lesion shrinkage and time to maximal shrinkage were not different between the IT and CT groups for lesions with volumes < 500 mm3. However, in lesions with volume > 500 mm3, 90% of lesions shrank after radiosurgery in the IT group compared with 47.8% in the CT group (p = 0.001). Median times to initial response and times to maximal shrinkage were faster in the IT group than in the CT group: initial response 49 days (95% CI 33.7-64.3 days) versus 84 days (95% CI 28.1-140 days), p = 0.001; maximal response 105 days (95% CI 59-150 days) versus 182 days (95% CI 119.6-244 days), p = 0.12., Conclusions: Unlike patients with melanoma, patients with NSCLC with brain metastases undergoing SRS showed no significant benefit-either in terms of survival or total amount of lesional response-when anti-PD-1 therapies were used. However, in lesions with volume > 500 mm3, combining SRS with IT may result in a faster and better volumetric response which may be particularly beneficial in lesions causing mass effect or located in neurologically critical locations.

Purpose: Treatment-related imaging changes (TRICs) after stereotactic radiosurgery (SRS) involves the benign transient enlargement of radiographic lesions after treatment. Identifying the radiation dose volumes and conformality metrics associated with TRICs for different post-treatment periods would be helpful and improve clinical decision making., Methods: 367 metastases in 113 patients were treated using Gamma Knife SRS between 1/1/2007-12/31/2009. Each metastasis was measured at each imaging follow-up to detect TRICs (defined as ≥ 20% increase in volume). Fluctuations in small volume lesions (less than 108 mm 3 ) were ignored given widely variable conformity indices (CI) for small volumes. The Karolinska Adverse Radiation Effect (KARE) factor, Paddick's CI, Shaw's CI, tumor volume (TV), 10 Gy (V10) and 12 Gy (V12) volumes, and prescription isodose volume (PIV) were calculated., Results: From 0-6 months, all measures correlated with the incidence of TRICs (p<.001), except KARE, which was inversely correlated. During the 6-12 month period all measures except KARE were still correlated. Beyond 12 months, no correlation was found between any of the measures and the development of TRICs., Conclusions: All metrics except KARE were associated with TRICs from 0-12 months only. Additional patient and treatment factors may become dominant at greater times after SRS., Competing Interests: Authors’ disclosure of potential conflicts of interest Drs. Chiang, Knisely, Qian, and Taylor have nothing to disclose. Dr. Yu reports grants from 21st Century Oncology during the conduct of the study.