Timothy Sum Hon Mun^1,2, Imogen Thrussell^1,2, Jessica Winfield^1,2, Amani Arthur³, David J Collins¹, Dow-Mu Koh¹, Paul Huang³, Simon J Doran¹, Christina Messiou¹, and Matthew D Blackledge^1
1Division of Radiotherapy and Imaging, Institute of Cancer Research, London, United Kingdom, ²Department of Radiology, Royal Marsden NHS Foundation Trust, London, United Kingdom, ³Institute of Cancer Research, London, United Kingdom

Monitoring treatment response of soft-tissue sarcomas (STS) following radiotherapy is challenging due to the inherent intratumoral heterogeneity of the disease. Radiomics and deep-learning provide opportunities for the discovery of potent biomarkers of treatment response. Successful response biomarkers must demonstrate good baseline repeatability if they are to be used for personalized treatment. We explore the stability of radiomic features derived from a deep-learning pipeline by determining the pairwise correlation of derived features, and measuring the baseline repeatability of features derived from the Apparent Diffusion-Coefficient maps. We demonstrate that 81/512 features are both independent and stable at repeat baseline measurement.  

Samuel Bobholz¹, Allison Lowman², Michael Brehler², Savannah Duenweg¹, John Sherman¹, Fitzgerald Kyereme², Elizabeth Cochran³, Dylan Coss³, Jennifer Connelly⁴, Wade Mueller⁵, Mohit Agarwal², Anjishnu Banerjee⁶, and Peter LaViolette^2,7
1Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ²Radiology, Medical College of Wisconsin, Milwaukee, WI, United States, ³Pathology, Medical College of Wisconsin, Milwaukee, WI, United States, ⁴Neurology, Medical College of Wisconsin, Milwaukee, WI, United States, ⁵Neurosurgery, Medical College of Wisconsin, Milwaukee, WI, United States, ⁶Biostatistics, Medical College of Wisconsin, Milwaukee, WI, United States, ⁷Biomedical Engineering, Medical College of Wisconsin, Milwaukee, WI, United States

This study used autopsy tissue samples to develop multi-stage radio-pathomic models of tumor probability in glioma patients. Three models were trained to predict cell density, extracellular fluid density, and cytoplasm density segmented from autopsy samples using T1, T1C, FLAIR, and ADC intensity. A fourth model was then trained to predict tumor probability from pathological annotations using the cellularity, extracellular fluid, and cytoplasm segmentations as input. The combined models were then able to non-invasively estimate tumor probability using MRI. These maps identified regions of tumor beyond the contrast-enhancing region and discriminated between areas of tumor and vasogenic edema within FLAIR hyperintensity.

Yoshiharu Ohno^1,2, Masao Yui³, Kaori Yamamoto³, Daisuke Takenaka⁴, Takeshi Yoshikawa⁴, Masato Ikedo³, Saki Takeda⁵, Akiyoshi Iwase⁵, Yuka Oshima¹, Nayu Hamabuchi¹, Satomu Hanamatsu¹, Yuki Obama¹, Hiroyuki Nagata¹, Takahiro Ueda¹, Hirotaka Ikeda¹, Kazuhiro Murayama², and Hiroshi Toyama^1
1Radiology, Fujita Health University School of Medicine, Toyoake, Japan, ²Joint Research Laboratory of Advanced Medical Imaging, Fujita Health University School of Medicine, Toyoake, Japan, ³Canon Medical Systems Corporation, Otawara, Japan, ⁴Diagnostic Radiology, Hyogo Cancer Center, Akashi, Japan, ⁵Radiology, Fujita Health University Hospital, Toyoake, Japan

No report has been found to compare nodule detection and Lung RADS classification capabilities in lung cancer screening cohort among pulmonary MR imaging with UTE, low-dose CT (LDCT) and standard-dose CT (SDCT).  We hypothesized that pulmonary MR imaging with UTE has a similar potential to detect pulmonary nodules and evaluate Lung-RADS classification and can apply lung cancer screening as well as CT.  The purpose of this study was to compare the capability for nodule detection and Lung RADS classification among pulmonary MR imaging with UTE, LDCT and SDCT in lung cancer screening population. 

Xue Ren¹, Jiazheng Wang², Liangjie Lin², Qingwei Song¹, Renwang Pu¹, Ying Zhao¹, Tao Lin¹, Qihao Xu¹, Zhiwei Shen², and Ailian Liu^1
1Department of Radiology, the First Affiliated Hospital of Dalian Medical University, Dalian, China, ²Philips Healthcare, Beijing, China

This study aims to assess the efficacy of amide proton transfer-weighted (APTw) combined with intravoxel incoherent motion (IVIM) imaging in differentiation  benign liver lesions, primary malignant and secondary malignant tumors. Results showed that a high diagnostic efficacy could be achieved through the combination use of the APT value and IVIM parameters.

Michal R Tomaszewski¹, Kujtim Latifi², Emanuel Boyer³, Russell F Palm³, Issam El Naqa⁴, Eduardo G Moros², Sarah E Hoffe³, Stephen A Rosenberg³, Jessica M Frakes³, and Robert J Gillies^1
1Cancer Physiology, H Lee Moffitt Cancer Center, Tampa, FL, United States, ²Medical Physics, H Lee Moffitt Cancer Center, Tampa, FL, United States, ³Radiation Oncology, H Lee Moffitt Cancer Center, Tampa, FL, United States, ⁴Machine Learning, H Lee Moffitt Cancer Center, Tampa, FL, United States

Magnetic Resonance Image guided Stereotactic body radiotherapy (MRgRT) is increasingly used in treatment of multiple cancers including pancreatic adenocarcinoma (PDAC). We hypothesized that quantitative analysis (radiomics) of the longitudinal MRgRT imaging during treatment can help predict response. MRgRT TrueFISP images from n=26 non-resectable PDAC patients were analyzed and image feature ratios last/first fraction quantified. Image normalization to kidney signal was validated and robustness of features assessed. Histogram skewness change demonstrated significant association with Progression Free Survival. This result shows promise for future application of the novel integrated framework for processing and quantification of MRgRT data presented here first time.

Savannah R. Duenweg¹, Samuel A Bobholz², Allison K Lowman³, Michael Brehler³, Fitzgerald Kyereme³, Kenneth A Iczkowski⁴, and Peter S LaViolette^3,5
1Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ²Biophysics, Medical College of Wisconsin, Wauwatosa, WI, United States, ³Radiology, Medical College of Wisconsin, Wauwatosa, WI, United States, ⁴Pathology, Medical College of Wisconsin, Wauwatosa, WI, United States, ⁵Biomedical Engineering, Medical College of Wisconsin, Wauwatosa, WI, United States

This study used digitized histology and MP-MRI from 48 patients with prostate cancer (PCa) to determine the relationship between complex histomorphometric features of PCa histology and MRI features. After prostatectomy, tissue samples were digitized, and co-registered to the T2 image.  Slides were annotated by a pathologist, individual glands were identified using automated image processing algorithms, and histomorphometric features were calculated. Radiomic features were calculated to compare higher order texture and histological features. We found that lumen was more strongly associated with raw intensity, whereas epithelial features were more strongly associated with computed texture features.

Lars Johannes Isaksson¹, Paul E Summers², Matteo Johannes Pepa¹, Mattia Zaffaroni¹, Maria Giulia Vincini¹, Giulia Corrao^1,3, Giovanni Carlo Mazzola^1,3, Marco Rotondi^1,3, Sara Raimondi⁴, Sara Gandini⁴, Stefania Volpe^1,3, Zaharudin Haron⁵, Sarah Alessi², Paola Pricolo², Francesco Alessandro Mistretta⁶, Stefano Luzzago⁶, Federico Cattani⁷, Gennaro Musi^3,6, Ottavio De Cobelli^3,6, Marta Cremonesi⁸, Roberto Orecchia⁹, Giulia Marvaso^1,3, Barbara Alicja Jereczek-Fossa^1,3, and Giuseppe Petralia^3,10
1Division of Radiation Oncology, IEO, European Institute of Oncology IRCCS, Milano, Italy, ²Division of Radiology, IEO, European Institute of Oncology IRCCS, Milano, Italy, ³Department of Oncology and Hemato-oncology, University of Milan, Milano, Italy, ⁴Department of Experimental Oncology, IEO, European Institute of Oncology IRCCS, Milano, Italy, ⁵Radiology Department, National Cancer Institute, Putrajaya, Malaysia, ⁶Division of Urology, IEO, European Institute of Oncology IRCCS, Milano, Italy, ⁷Unit of Medical Physics, IEO, European Institute of Oncology IRCCS, Milano, Italy, ⁸Radiation Research Unit, IEO, European Institute of Oncology IRCCS, Milano, Italy, ⁹Scientific Directorate, IEO, European Institute of Oncology IRCCS, Milano, Italy, ¹⁰Precision Imaging and Research Unit, IEO, European Institute of Oncology IRCCS, Milano, Italy

The risk of patients being under- or overtreated during radiotherapy depends heavily on the pre-treatment assessment. Prediction models for surgical margin status, pathological lymph nodes, pathological tumor stage and ISUP grade group were formed using clinical and radiological features alone and together with whole-prostate radiomic features in 100 patients who proceeded to prostatectomy after multiparametric-MRI. The addition of radiomics features significantly improved AUC for the prediction models. The leading radiomic features differed between the different models.

Mohammad Alhusseini¹, Adrian L Breto¹, Isaac L Xu¹, Ahmad Algohari¹, Sandra M Gaston¹, Matthew C Abramowitz¹, Alan Dal Para¹, Sanoj Punnen², Alan Pollack¹, and Radka Stoyanova^1
1Department of Radiation Oncology, University of Miami Miller School of Medicine, Miami, FL, United States, ²Department of Urology, University of Miami Miller School of Medicine, Miami, FL, United States

Circulating tumor cells (CTCs) have been shown to be an indicator for metastatic risk in prostate cancer. We investigated the association between radiomics features extracted from multiparametric MRI of the prostate and CTC counts in prostate cancer patients enrolled in two institutional clinical trials (n=71). We trained a neural network to predict the dichotomized CTCs counts, defined by a 5 CTCs threshold. The top seven features, ranked using maximum-relevance minimum-redundancy, were used as input to a neural network. The training and testing were repeated for 100 runs of 5-Fold cross validation, resulting in AUC 0.834 to predict CTCs ≥ 5.