Joseph Holtrop¹, John R Glass¹, Stuart S McAfee¹, Silu Zhang¹, Matthew A Scoggins¹, Amar Gajjar², Giles W Robinson², Wilbur E Reddick¹, and Asim K Bag^1
1Department of Diagnostic Imaging, St. Jude Children's Research Hospital, Memphis, TN, United States, ²Department of Pediatrics, St. Jude Children's Research Hospital, Memphis, TN, United States

Survivors of medulloblastoma patients frequently have cognitive impairment, which is currently thought to be due to radiation and chemotherapy. In this work, we have showed that there is widespread changes in diffusion tensor parameters in cerebral white and gray matters, suggestive of brain microstructure damage, even before start of radiation and chemotherapy and not related to hydrocephalus.  

Ina Ly¹, Yiqiao Song¹, Qiuyun Fan², Aapo Nummenmaa¹, Maria Martinez-Lage Alvarez¹, William Curry¹, Brian Nahed¹, Daniel Cahill¹, Pamela Jones¹, Jorg Dietrich¹, Deborah Forst¹, Scott Plotkin¹, Tracy Batchelor³, Bruce Rosen¹, Susie Huang¹, and Elizabeth Gerstner^1
1Massachusetts General Hospital, Boston, MA, United States, ²Department of Biomedical Engineering, College of Precision Instruments and Optoelectronics Engineering, Tianjin University, Tianjin, China, ³Brigham and Women's Hospital, Boston, MA, United States

Diffuse gliomas demonstrate significant spatial heterogeneity which is not captured on standard anatomical MRI or with existing diffusion MRI (dMRI) methods. This is because current dMRI approaches provide voxel-averaged information about tissue composition and are based on a priori assumptions about underlying tissue microstructure, thus providing a simplified model of tissue properties. Here, we apply an unbiased, model-agnostic dMRI method (model-free diffusion tensor distribution (FDTD) and K-means clustering) to seven subjects with diffuse gliomas. We confirm the presence of intratumoral heterogeneity and find distinct differences in diffusion properties between gliomas of different histologic grades and pre- and post-treatment.

Deepti Upadhyay¹, Jinny Sun¹, Shubhangi Agarwal¹, Robert Bok¹, Romelyn DeLos Santos¹, Donna Peehl¹, John Kurhanewicz¹, and Renuka Sriram^1
1Department of Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States

NMR-based stable isotope resolved metabolomics was used to characterize small cell neuroendocrine (SCNC) and adenocarcinoma (aCRPC) subtypes of castration-resistant prostate cancer patient-derived xenografts (PDXs). Targeted metabolomics indicated distinct upregulation of metabolic pathways in SCNC relative to aCRPC PDXs. Specifically, [U-¹³C]glucose and [U-¹³C]glutamine labeling demonstrated that SCNC PDXs had increased glycolytic rate, alanine aminotransferase and tricarboxylic acid cycle activity, including anaplerotic sources. Further, the metabolic differences observed among the aCRPC and SCNC PDXs exhibited a continuous range as predicted by clinical genomic data of these subtypes of CRPC.

Victor D. Schepkin¹, Cathy W. Levenson², Shannon Helsper³, and Dean Sherry^4
1CIMAR, NHMFL/FSU, Tallahassee, FL, United States, ²College of Medicine, FSU, Tallahassee, FL, United States, ³NHMFL/FSU, Tallahassee, FL, United States, ⁴Southwestern Medical Center, UT Dallas, Dallas, TX, United States

Labeled glucose[6-¹⁷O] was used to evaluate differences in glucose metabolism between normal and tumor brain tissue. Experiments were performed using an intracranial 9L glioma model at 21.1 T magnet by observing changes of ¹⁷O-water and ¹⁷O-glucose MR peaks.  The estimated rate of the total glucose metabolism in 9L glioma was approximately 11 times more than in the normal brain.   Correspondingly, glucose metabolism rate is 7.5 times more in glioma at the enolase step. ¹⁷O MRI/MRS is an encouraging tool for assessing the differences in glucose metabolism in various cancer cells.

 

Khadija Sheikh¹, Junghoon Lee², Marc Morcos², Mohammad Rezae², Ravi T. Seethamraju³, Thomas Benkert⁴, Himanshu Bhat³, Henry R Halperin⁵, Bruce L. Daniel⁶, Akila N Viswanathan², and Ehud J Schmidt^2,5
1Johns Hopkins School of Medicine, Washington, DC, United States, ²Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins School of Medicine, Baltimore, MD, United States, ³Siemens Medical Solutions, Boston, MA, United States, ⁴MR Applications Predevelopment, Siemens Healthcare GmbH, Erlangen, Germany, ⁵Cardiology, Johns Hopkins School of Medicine, Baltimore, MD, United States, ⁶Radiology, Stanford University, Stanford, CA, United States

Imaging the temporal evolution of post-radiation fibrosis, which forms in irradiated tissues, supports quantifying tumor response and estimating surrounding-tissue injury.  Five cervical cancer patients were scanned pre-external-beam-radiotherapy (EBRT), during and after EBRT, and 3-months post-EBRT and High-Dose-Rate (HDR) brachytherapy.   A novel stack-of-spirals STIR-UTE sequence (TI=60ms) was used to image acute-fibrosis, utilizing collagen-bound-water’s short T2* and T1, and Late-Gadolinium-Enhancement (11-15min post-contrast) IR-UTE sequence (TI=200ms) to image chronic-fibrosis, utilizing scar’s slow-contrast-perfusion. Acute-fibrosis spatial-distribution followed EBRT and brachytherapy dose distributions. Over time, acute-fibrosis was converted into chronic-fibrosis.   This is important as it forms the basis to guide radiation-therapy for re-irradiation and adaptive-planning.

Ante Zhu¹, Robert Y. Shih^2,3, James Kevin DeMarco^2,3, Herman Douglas Morris^2,3, Radhika Madhavan¹, Gail Kohls^2,3, Tim Sprenger⁴, Maureen Hood^2,3, Luca Marinelli¹, Jason A. Gregory^2,3, Brett J. Theeler^2,3, Vincent B. Ho^2,3, and Thomas K.F. Foo^1
1Biology and Applied Physics, GE Research, Niskayuna, NY, United States, ²Uniformed Services University of the Health Sciences, Bethesda, MD, United States, ³Walter Reed National Military Medical Center, Bethesda, MD, United States, ⁴GE Healthcare, Stockholm, Sweden

Time-dependent diffusion MRI has shown high specificity to non-Gaussian restricted diffusion, which improves assessing cytoarchitectural alternations in pre-/post- treatment glioma. However, the characterization of time-dependent diffusivity ADC(f) and kurtosis MK(f) at short diffusion time using oscillating gradient spin echo (OGSE) diffusion encoding is limited. In our preliminary studies at ultra-high-gradient 3.0T, larger changes of ADC(f) and MK(f) were shown in glioblastoma compared to a low-grade astrocytoma, indicating highly restricted microenvironment as confirmed on histopathology. OGSE-based time-dependent diffusion showed promise for evaluating cytoarchitectural alternations in gliomas. ADC(f) and MK(f) of contrast-enhancing lesions in three post-glioblastoma-treatment patients showed strong time-dependence.

Zhibo Zhu¹, Yannick Bliesener¹, Frances Chow², Jay Acharya³, and Krishna S. Nayak^4
1Department Of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, United States, ²Department of Neurosurgery, University of Southern California, Los Angeles, CA, United States, ³Department of Radiology, University of Southern California, Los Angeles, CA, United States, ⁴Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, United States

We perform a prospective test-retest repeatability study of one such approach, in patients with high-grade glioma. Two separate scans were performed within 3-7 days apart. Tumor, scalp, normal white matter, pons, cerebellum and muscle regions were segmented manually. The coefficients of variation of median Kt were 50.4% to 66.6% and of median vp were 23.2% to 80.0%, based on our first 5 subjects.

Manjunathan Nanjappa¹ and Arunark Kolipaka^1
1Department of Radiology, The Ohio State University, Columbus, OH, United States

Magnetic Resonance Elastography is an emerging alternative diagnostic tool for palpation and it is known that stiffness is altered in chronic pancreatitis and as well as in pancreatic tumors. In this study we compared performances of a drum-shaped rigid driver with a flat soft-pad passive driver in 10 volunteers at 40Hz and 60Hz diver frequencies. The variations of mean stiffness values measured between the drivers were found to be insignificant at a given frequency. Additionally, stiffness measured at 40Hz is lower compared to 60Hz. Therefore, users can make their choice of the driver based on patient comfort, longevity and cost.

Avishkar Sharma¹, Ke Lei², Shreyas Vasanawala¹, and Vipul Sheth^1
1Radiology, Stanford University, Stanford, CA, United States, ²Electric Engineering, Stanford University, Stanford, CA, United States

Magnetic resonance (MR) imaging plays a pivotal role in the staging and treatment planning of rectal cancer. Accurate staging depends on good-quality high-resolution axial T2-weighted images orthogonal to the rectal tumor. Rectal MRI is often confounded by motion artifacts secondary to bowel peristalsis and patient movement. We propose a CNN model that automatically assesses image quality instantaneously after a scan is finished to reduce the frequency of patient recalls and non-diagnostic images. Our model achieves high accuracy in identifying motion degradation on an individual slice basis and perfect accuracy when classifying the entire sequence.  

Yang Zhang^1,2, Farouk Nouizi¹, Ning Lang³, Xiaoying Xing³, Yongye Chen³, Qizheng Wang³, Enlong Zhang³, Huishu Yuan³, Ke Nie², and Min-Ying Su^1
1Department of Radiological Sciences, University of California, Irvine, CA, United States, ²Department of Radiation Oncology, Rutgers-Cancer Institute of New Jersey, Rutgers-Robert Wood Johnson Medical School, New Brunswick, NJ, United States, ³Department of Radiology, Peking University Third Hospital, Beijing, China

A total of 27 patients receiving both spinal CT and MR for evaluation of back pain were identified for analysis. MR images and CT image were co-registered first, and the CT was used as ground truth for training a deep learning algorithm using MR images to generate synthetic CT. In this study, we implemented cycleGAN to generate these synthetic CT images from their corresponding MR slices.  Five-fold cross validation was used to evaluate the performance of the trained model.  Compared to the original images, the Mean Average Error was 27.63±11.51, and the Peak Signal-to Noise Ratio was 19.44±5.72.

Brigid McDonald¹, Renjie He¹, Yao Ding², Travis Salzillo², Jihong Wang², Clifton Fuller², and Abdallah Mohamed^2
1Radiation Oncology, UT MD Anderson Cancer Center, Houston, TX, United States, ²UT MD Anderson Cancer Center, Houston, TX, United States

In this study, we investigate weekly changes in mean and median apparent diffusion coefficient (ADC) values from diffusion-weighted MRI for HPV+ oropharyngeal cancer patients undergoing definitive radiation therapy (RT). Trends of increasing ADC values in GTV_P and GTV_N were observed with clear separation at each time point between patients with no evidence of disease vs. recurrence at 2 months post-RT. Because of our small sample size (15), a larger cohort is needed for more robust outcome prediction analysis.

Jie Deng¹, Chenyang Shen¹, Junjie Wu¹, and Steve B Jiang^1
1Radiation Oncology, UT Southwestern Medical Center, Dallas, TX, United States

MRI-only simulation permits superior soft tissue contrast for tumor tissue characterization and target margin delineation. One bottleneck preventing MRI-only simulation to be widely applied in prostate cancer EBRT is the tedious marker localization process on MR images as compared to that on a CT.  The fiducial and calcification on MRI also appear to be similar, which reduces the confidence level of the clinician in fiducial localization, and further increases the time needed. In this study, we developed a framework using quantitative susceptibility mapping to automatically identify the fiducial markers based on phase information of gradient-echo MRI.

Yang Zhang^1,2, Liming Shi³, Weiwen Zhou³, Xiaonan Sun³, Salma Jabbour¹, Ning Yue¹, Min-Ying Su², and Ke Nie^1
1Department of Radiation Oncology, Rutgers-Cancer Institute of New Jersey, Rutgers-Robert Wood Johnson Medical School, New Brunswick, NJ, United States, ²Department of Radiological Sciences, University of California, Irvine, CA, United States, ³Department of Radiation Oncology, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China

A radiomics model with pre-treatment MRI for pCR prediction and compared with two oncologists’ readings. Their performance with and without the model assistances was cross-compared to see the potential role of radiomics model in assisting clinical decision. A total of 203 patients receiving neoadjuvant CRT followed by total mesorectal excision (TME) were enrolled. For training set, AUC from radiomics model is 0.85 and AUC from clinical model is 0.74. For testing set, AUC from radiomics model is 0.82 and AUC from clinical model is 0.69. For oncologist’s reading, with the assistance of radiomics model, positive prediction value (PPV) and specificity increased significantly for both experienced and inexperienced oncologist.

Harshan Ravi¹, Andres M. Arias Lorza¹, James R. Costello², Hyo Sook Han³, Daniel K. Jeong², Stephan G. Klinz Klinz⁴, Jasgit C. Sachdev⁵, Ronald L. Korn⁶, and Natarajan Raghunand^1,7
1Department of Cancer Physiology, Moffitt Cancer Center and Research Institute, Tampa,, FL, United States, ²Department of Radiology, Moffitt Cancer Center and Research Institute, Tampa,, FL, United States, ³Department of Breast Oncology, Moffitt Cancer Center and Research Institute, Tampa,, FL, United States, ⁴Ipsen Bioscience, Cambridge, MA, United States, ⁵HonorHealth Research Institute, Scottsdale, AZ, United States, ⁶Imaging Endpoints Core Lab, Scottsdale, AZ, United States, ⁷Department of Oncologic Sciences, University of South Florida, Tampa,, FL, United States

  R^*₂ and the apparent concentration of FMX in the tumor (FMX_C) are considered  biomarkers of liposomal irinotecan (nal-IRI) drug uptake into that tumor lesion, which in turn would determine response of that tumor to nal-IRI. Historically, quantification of R^*₂ and FMX_C  had been implemented by using a calibration phantom and acquiring patient scans both pre-FMX and post-FMX. Here we have demonstrated a pre-treatment FMX-enhanced MRI companion biomarker of response to nal-IRI in mBC patients that can be computed from a single16-24 h post-FMX MRI scan without the need for calibration phantoms or pre-FMX scans.

Harshan Ravi¹, Andres M. Arias Lorza¹, James R. Costello², Hyo Sook Han³, Daniel K. Jeong², Stephan G. Klinz⁴, Jasgit C. Sachdev⁵, Ronald L. Korn⁶, and Natarajan Raghunand^1,7
1Department of Cancer Physiology, Moffitti Cancer Center and Research Institute, Tampa,, FL, United States, ²Department of Radiology, Moffitti Cancer Center and Research Institute, Tampa, FL, United States, ³Department of Breast Oncology, Moffitti Cancer Center and Research Institute, Tampa, FL, United States, ⁴Ipsen Bioscience,, Cambridge,, MA, United States, ⁵Honor Health Research Institute, Scottsdale,, AZ, United States, ⁶Imaging Endpoints Core Lab, Scottsdale,, AZ, United States, ⁷Department of Oncologic Sciences, University of South Florida, Tampa, FL, United States

Phagocytosis of ferumoxytol (FMX) by tumor-associated macrophages (TAMs) results in intracellular compartmentation, which has opposing effects on longitudinal (R₁) and transverse (R₂^*) relaxivity of FMX. Pixelwise mismatch between apparent ferumoxytol concentration (FMX_c) computed from  post-FMX R₁ and R₂^* maps can be exploited to identify pixels with high concentrations of phagocytosed (compartmentalized) FMX. Here we show mismatch on MRI, acquired 24 h post-FMX, measured in terms of compartmentation index (CI) greater than 1 and hypothesized to be indicative of FMX intracellular compartmentation, was predictive of the response of breast cancer brain metastases to liposomal irinotecan (nal-IRI) at the individual tumor level.

Katie Hanna¹, Ehab Husain², Yazan Masannat³, Rasha Abu-Eid⁴, David Lurie⁵, Valerie Speirs¹, and Lionel Broche^5
1Institute of Medical Sciences, School of Medicine, Medical Sciences and Nutrition, University of Aberdeen, Aberdeen, Scotland, ²Department of Pathology, Aberdeen Royal Infirmary, Aberdeen, Scotland, ³Breast Unit, Aberdeen Royal Infirmary, Aberdeen, Scotland, ⁴Institute of Dentistry, School of Medicine, Medical Sciences and Nutrition, University of Aberdeen, Aberdeen, Scotland, ⁵Aberdeen Biomedical Imaging Centre, University of Aberdeen, Aberdeen, Scotland

This study aims to explore novel biomarkers of breast cancer using FFC-NMR. NMR dispersion profiles were generated by analysing patient-matched tumour, peritumoral zone and distant normal fixed tissue samples from twenty breast cancer patients. These profiles were analysed using piecewise power models and the numerical parameters derived from these models could significantly classify the different regions of breast tissue and distinguish patients from different prognostic categories. The numerical parameters investigated in this study may have potential as quantitative biomarkers for breast cancer detection and risk stratification. 

Davide Prezzi¹, Radhouene Neji^1,2, Isabel Dregely¹, Sami Jeljeli¹, Paul Bassett³, Gary Cook¹, and Vicky Goh^1
1King's College London, London, United Kingdom, ²Siemens Healthineers, Frimley, United Kingdom, ³Statsconsultancy Ltd, Amersham, United Kingdom

Oxygen-enhanced ΔT1 tumor measurements were feasible in 12/22 colorectal cancer patients by means of single-slice MOLLI. A small but statistically significant negative tumor ΔT1 was found overall, in line with the expected response in well perfused and oxygenated tissue [mean (95% CI) ΔT1 =  -33 (-41, -25) msec; mean tumor T1 = 1545 ± 163 msec]. A significant negative MOLLI ΔT1 was found in muscle. Repeatability and interobserver agreement were acceptable. Multi-slice variable-flip-angle VIBE tumor T1 measurements were degraded by motion or susceptibility artefact. Technical developments aimed at mitigating artefact and amplifying ΔT1 are needed for future clinical translation.