Purpose: In an effort of early assessment of treatment response, we investigate radiation induced changes in CT number histogram of GTV during the delivery of chemoradiation therapy (CRT) for pancreatic cancer. Methods: Diagnostic‐quality CT data acquired daily during routine CT‐guided CRT using a CT‐on‐rails for 20 pancreatic head cancer patients were analyzed. All patients were treated with a radiation dose of 50.4 in 28 fractions. On each daily CT set, the contours of the pancreatic head and the spinal cord were delineated. The Hounsfiled Units (HU) histogram in these contourswere extracted and processed using MATLAB. Eight parameters of the histogram including the mean HU over all the voxels, peak position, volume, standard deviation (SD), skewness, kurtosis, energy, and entropy were calculated for each fraction. The significances were inspected using paired two‐tailed t‐test and the correlations were analyzed using Spearman rank correlation tests. Results: In general, HU histogram in pancreatic head (but not in spinal cord) changed during the CRT delivery. Changes from the first to the last fraction in mean HU in pancreatic head ranged from −13.4 to 3.7 HU with an average of −4.4 HU, which was significant (P<0.001). Among other quantities, the volume decreased, the skewness increased (less skewed), and the kurtosis decreased (less sharp) during the CRT delivery. The changes of mean HU, volume, skewness, and kurtosis became significant after two weeks of treatment. Patient pathological response status is associated with the changes of SD (ΔSD), i.e., ΔSD= 1.85 (average of 7 patients) for good reponse, −0.08 (average of 6 patients) for moderate and poor response. Conclusion: Significant changes in HU histogram and the histogram‐based metrics (e.g., meam HU, skewness, and kurtosis) in tumor were observed during the course of chemoradiation therapy for pancreas cancer. These changes may be potentially used for early assessment of treatment response.

BackgroundAccurate delineation of the gross tumor volumes (GTV) is a prerequisite for precise radiotherapy planning and delivery. Different MRI sequences have different advantages and limitations in their ability to discriminate primary cervical tumor from normal tissue. The purpose of this work is to determine appropriate MRI techniques for GTV delineation for external-beam radiation therapy of locally advanced cervical cancer (LACC).Materials and MethodsGTVs were delineated on the MRI, CT, and PET images acquired for 23 LACC patients in treatment positions to obtain GTVs on CT (GTV-CT), on various MRI sequences including T1 (GTV-T1), T2 (GTV-T2), T1 with fat suppression and contrast (GTV-T1F+), DWI-ADC (GTV-ADC) and on PET were generated using the threshold of 40% of maximum SUV (GTV-SUV40%) as well as SUV of 2.5 (GTV-SUV2.5). MRI, CT and PET were registered for comparison. The GTVs defined by MRI were compared using the overlap ratio (OR) and relative volume ratio (RVR). The union of GTV-T2 and GTV-ADC was generated to represent the MRI-based GTV (GTV-MRI).ResultsThe differences between GTV-T2 and other MRI GTVs are significant (P < 0.05). The average ORs for GTV-T1, GTV-T1F+, and GTV-ADC related to GTV-T2 were 86.3%, 81.6%, and 61.6% with the corresponding average RVRs 113.8%, 112.3% and 77.2%, respectively. There is no significant difference between GTV-T1 and GTV-T1F+. GTV-ADC was generally smaller than GTV-T2, however, encompassed suspicious regions that are uncovered in GTV-T2 (up to 16% of GTV-T2) because of different imaging mechanisms. There was significant difference between GTV-MRI, GTV-SUV2.5, GTV-SUV40%, and GTV-CT. On average, GTV-MRI is 18.4% smaller than GTV-CT.ConclusionsMRI provides improved visualization of disease over CT or PET for cervical cancer. The GTV from the union of GTV-T2 and GTV-ADC provides a reasonable GTV including tumor region defined anatomically and functionally with MRI and substantially reduces the conventional GTV defined on CT.

Purpose: Acute hematologic toxicity associated with bone marrow injury is a common complication of chemoradiation therapy (CRT) for pelvic malignancies. In this work, we investigate the feasibility of using quantitative CT to detect bone marrow injury during CRT. Methods: Daily CTs were acquired during routine CT‐guided radiation therapy using a CT‐on‐rails for 15 cervical cancer patients. All patients treated with a radiation dose of 45.0 to 50.4 Gy in 1.8 Gy/fraction along with chemotherapy. For each patient, the contours of bone marrow were generated in L4, L5 and sacrum on the first daily CT and then populated to other daily CTs by rigid registration using MIM (MIM Software Inc., Cleveland, OH) with manual editing if possible. A series of CT texture parameters, including Hunsfield Unit (HU) histogram, mean HU, entropy, energy, in bone marrow contours were calculated using MATLAB on each daily CT and were correlated with the completed blood counts (CBC) collected weekly for each patient. The correlations were analyzed with Pearson correlation tests. Results: For all patient data analyzed, mean HU in bone marrow decreased during CRT delivery. From the first to the last fraction the average mean HU reduction is 58.1 ± 13.6 HU (P<0.01). This decrease can be observed as early as after first 5 fractions and is strongly associated with the changes of most CBC quantities, such as the reductions of white and blood cell counts (r=0.97, P=0.001). The reduction of HU is spatially varied. Conclusion: Chemoradiation induced bone marrow injury can be detected during the delivery of CRT using quantitative CT. Chemoradiation results in reductions in mean HU, which are strongly associated with the change in the pretrial blood cell counts. Early detection of bone marrow injury with commonly available CT opens a door to improve bone marrow sparing, reducing risk of hematologic toxicity.

Purpose: To quantify interfractional anatomic changes and the dosimetric variations in the parotid and submandibular glands in image‐guided IMRT for head and neck cancer. Methods: Daily diagnostic‐quality CTs acquired during daily IGRT using an in‐room CT (CTVision, Siemens) for 16 head and neck cancer patients treated with IMRT were analyzed. The contours of parotid glands and submandibular glands on dialy CTs were generated using an auto‐segmentation tool (ABAS, Elekta) with manual editing by a single physician. The volumetric changes were measured and the center of mass (COM) of each gland was obtained. The positional changes of the glands were characterized by inter‐parotid and inter‐submandibular gland distances. The original plan was applied to each fraction CT with isocenter shifts based on the alignment of the tumor. The DVHs and commonly used dose volume parameters from daily plans were compared to the original plans. Results: From fraction 1 to fraction 31, the average volume reduction was roughly 25% for all glands. The changes of inter‐parotid and inter‐submandibular gland distances were significant, with these distances decreased by 1.7±4.3 mm and 1.4±2.7 mm in average from fraction 1 to fraction 31. The average distance changes between the glands to the isocenter were from 0.4 to 3.3 mm. The mean dose changes for the four glands were from 109 cGy to 265 cGy. Conclusion: Significant decreases in the volume, inter‐gland distance were found during IG‐IMRT for head and neck cancer. These changes result in significant variations in doses to these glands, indicating the need of adaptive replanning.

Purpose: Apparent diffusion coefficient (ADC) map may help to delineate the gross tumor volume (GTV) in prostate gland. Dose painting with external beam radiotherapy for GTV might increase the local tumor control. The purpose of this study is to explore the maximum boosting dose on GTV using VMAT without sacrificing sparing of organs at risk (OARs) in MRI based planning. Methods: VMAT plans for 5 prostate patients were generated following the commonly used dose volume (DV) criteria based on structures contoured on T2 weighted MRI with bulk electron density assignment using electron densities derived from ICRU46. GTV for each patient was manually delineated based on ADC maps and fused to T2‐weighted image set for planning study. A research planning system with Monte Carlo dose engine (Monaco, Elekta) was used to generate the VMAT plans with boosting dose on GTV gradually increased from 85Gy to 100Gy. DV parameters, including V(boosting‐dose) (volume covered by boosting dose) for GTV, V75.6Gy for PTV, V45Gy, V70Gy, V72Gy and D1cc (Maximum dose to 1cc volume) for rectum and bladder, were used to measure plan quality. Results: All cases achieve at least 99.0% coverage of V(boosting‐dose) on GTV and 95% coverage of V75.6Gy to the PTV. All the DV criteria, V45Gy≤50% and V70Gy≤15% for bladder and rectum, D1cc ≤77Gy (Rectum) and ≤80Gy (Bladder), V72Gy≤5% (rectum and bladder) were maintained when boosting GTV to 95Gy for all cases studied. Except for two patients, all the criteria were also met when the boosting dose goes to 100Gy. Conclusion: It is dosimetrically feasible safe to boost the dose to at least 95Gy to ADC defined GTV in prostate cancer using MRI guided VMAT delivery. Conclusion: It is dosimetrically feasible safe to boost the dose to at least 95Gy to ADC defined GTV in prostate cancer using MRI guided VMAT delivery. This research is partially supported by Elekta Inc.