Automatic segmentation of magnetic resonance images for high-dose-rate cervical cancer brachytherapy using deep learning.

PURPOSE: Magnetic resonance (MR) imaging is the gold standard in image-guided brachytherapy (IGBT) due to its superior soft-tissue contrast for target and organs-at-risk (OARs) delineation. Accurate and fast segmentation of MR images are very important for high-quality IGBT treatment planning. The purpose of this work is to implement and evaluate deep learning (DL) models for the automatic segmentation of targets and OARs in MR image-based high-dose-rate (HDR) brachytherapy for cervical cancer. METHODS: A 2D DL model using residual neural network architecture (ResNet50) was developed to contour the targets (gross tumor volume (GTV), high-risk clinical target volume (HR CTV), and intermediate-risk clinical target volume (IR CTV)) and OARs (bladder, rectum, sigmoid, and small intestine) automatically on axial MR slices of HDR brachytherapy patients. Furthermore, two additional 2D DL models using sagittal and coronal images were also developed. A 2.5D model was generated by combining the outputs from axial, sagittal, and coronal DL models. Similarly, a 2D and 2.5D DL models were also generated for the inception residual neural network (InceptionResNetv2 (InRN)) architecture. The geometric (Dice similarity coefficient (DSCs) and 95th percentile of Hausdorff distance (HD)) and dosimetric accuracy of 2D (axial only) and 2.5D (axial + sagittal + coronal) DL model generated contours were calculated and compared. RESULTS: The mean (range) DSCs of ResNet50 across all contours were 0.674 (0.05-0.96) and 0.715 (0.26-0.96) for the 2D and 2.5D models, respectively. For InRN, these were 0.676 (0.11-0.96) and 0.723 (0.35-0.97) for the 2D and 2.5D models, respectively. The mean HD of ResNet50 across all contours was 15.6 mm (1.8-69 mm) and 12.1 mm (1.7-44 mm) for the 2D and 2.5D models, respectively. The similar results for InRN were 15.4 mm (2-68 mm) and 10.3 mm (2.7-39 mm) for the 2D and 2.5D models, respectively. The dosimetric parameters (D90) of GTV and HR CTV for manually contoured plans matched better with the 2.5D model (p > 0.6) and the results from the 2D model were slightly lower (p < 0.08). On the other hand, the IR CTV doses (D90) for all of the models were slightly lower (2D: -1.3 to -1.5 Gy and 2.5D: -0.5 to -0.6 Gy) and the differences were statistically significant for the 2D model (2D: p < 0.000002 and 2.5D: p > 0.06). In case of OARs, the 2.5D model segmentations resulted in closer dosimetry than 2D models (2D: p = 0.07-0.91 and 2.5D: p = 0.16-1.0). CONCLUSIONS: The 2.5D DL models outperformed their respective 2D models for the automatic contouring of targets and OARs in MR image-based HDR brachytherapy for cervical cancer. The InceptionResNetv2 model performed slightly better than ResNet50.

Should molecular subtype be recommended as one of the selection criteria for accelerated partial breast irradiation? Preliminary results from an Asian cohort.

PURPOSE: The purpose of this study was to report clinical outcomes in patients treated with accelerated partial breast irradiation (APBI), stratified as per molecular subtype and American Society for Therapeutic Radiology and Oncology/Groupe Européen de Curiethérapie and European Society for Radiotherapy & Oncology (ASTRO/GEC-ESTRO) patient selection criteria in order to determine whether molecular subtype should be recommended as one of the selection criteria for APBI. MATERIAL AND METHODS: 157 early-stage breast cancers patients, treated with APBI using multi-catheter interstitial brachytherapy with ≥ 6 months follow-up were included. Molecular subtype was assigned based on estrogen/progesterone receptor (ER/PR), Her2neu and tumor grade. Patients were stratified into ASTRO and GEC-ESTRO risk groups, as per updated ASTRO consensus statement (CS) and GEC-ESTRO recommendation, respectively. The Kaplan-Meier method was used to calculate the time to event data of clinical outcomes. RESULTS: With a median follow-up of 35 months, local control (LC) and locoregional control (LRC) were not significantly different among the different molecular subtypes (p = 0.19, p = 0.41, respectively). None of the APBI guidelines predicted risk of local or locoregional recurrence. Re-analyzing the data by replacing ER status with molecular subtype in the ASTRO-CS did not show any significant difference in LC/LRC across the various categories. Her2neu subtype was associated with significantly lower disease-free survival, cause specific survival, and overall survival than the luminal subtypes. CONCLUSIONS: None of the mentioned APBI guidelines predicted local or locoregional recurrence risk in our study population. Additional follow-up will be needed to recommend inclusion of molecular subtype (or at least HER2 receptor status) in the patient selection criteria for APBI.

Evaluation of quality indices during multifractionated pelvic interstitial brachytherapy for cervical cancer.

PURPOSE: To evaluate the impact of needle displacements on quality indices during multifractionated pelvic interstitial brachytherapy (IBT). METHODS AND MATERIALS: Patients undergoing boost IBT were included. Postprocedure planning and verification CT scans were obtained. Three-dimensional needle displacements were measured. Clinical target volume and organ at risk were delineated. Coverage index (CI), dose homogeneity index (DHI), dose nonuniformity ratio (DNR), V170, V200, V250, and dose received by 2cc of organs at risk were obtained at baseline. The displacements were simulated by shifting dwell positions, and dose point optimized and graphically optimized plans were generated. Wilcoxon test determined statistical significance of differences in quality indices. RESULTS: Fifteen patients were included and received five fractions of IBT over 3 days. Maximum displacements were observed in caudal direction (average, 19.1mm). At baseline, CI of 0.94 (range, 0.91-0.99), DHI of 0.90 (range, 0.80-0.94), and DNR of 0.10 (range, 0.05-0.10) were attained. The CI, DHI, and DNR in Day 3 dose point optimized plans were 0.76 (range, 0.4-0.99), 0.76 (range, 0.40-0.94), and 0.23 (range, 0.06-0.64), respectively. The difference in CI, DHI, and DNR between baseline and Day 3 dose point optimized plans was statistically significant (p = 0.002, 0.007, and 0.001, respectively). Day 3 graphically optimized plans were superior to Day 3 dose point optimized plans (CI, 0.82 vs. 0.76; p = 0.01). Graphically optimized could however improve CI without compromise in DHI, DNR, V170, V200, and V250 only in patients wherein caudal displacements ≤15mm. CONCLUSIONS: Caudal needle displacements during multifractionated IBT cause significant deterioration of quality indices. Replanning with graphically optimized and/or needle repositioning maybe required for maintaining the quality of IBT.

Can irradiation of potential cancer stem-cell niche in the subventricular zone influence survival in patients with newly diagnosed glioblastoma?

Glioblastoma progenitor or stem cells residing in the stem-cell niche in the subventricular zones (SVZ) can initiate or promote tumorigenesis. They can also migrate throughout the brain, resulting in disease progression. Irradiation of potential cancer stem-cell niche in the SVZ may influence survival. To analyze radiotherapy dose-volume parameters to the SVZ that correlate with survival in adequately treated patients with newly diagnosed glioblastoma, 40 adults with histopathologically proven supratentorial glioblastoma with available baseline imaging treated with postoperative conventionally fractionated focal conformal radiotherapy plus chemotherapy, available radiotherapy planning dataset, and documented event of progression or death or minimum 6-month follow-up were included in this retrospective study. Dose-volume parameters to the SVZ were extracted from treatment planning system and analyzed in relation to survival outcomes. Mean ipsilateral and contralateral SVZ volumes were 5.6 and 6.4 cc, respectively. With median follow-up of 15 months (interquartile range 12-18 months), median [95 % confidence interval (CI)] progression-free survival (PFS) and overall survival (OAS) was 11 months (95 % CI 8.9-13.0 months) and 17 months (95 % CI 11.6-22.4 months), respectively. Older age (>50 years), poor recursive partitioning analysis (RPA) class, and higher than median of mean contralateral SVZ dose were associated with significantly worse PFS and OAS. Multivariate analysis identified RPA class, Karnofsky performance status, and mean ipsilateral SVZ dose as independent predictors of survival. Increasing mean dose to the ipsilateral SVZ was associated with significantly improved OAS. Irradiation of potential cancer stem-cell niche influences survival outcomes in patients with newly diagnosed glioblastoma.

Quality assurance of multifractionated pelvic interstitial brachytherapy for postoperative recurrences of cervical cancers: a prospective study.

PURPOSE: To evaluate three-dimensional needle displacements during multifractionated interstitial brachytherapy (BT) for cervical cancers. METHODS AND MATERIALS: Patients scheduled to undergo pelvic interstitial BT for postoperative and or postradiation vault recurrences were included from November 2009 to December 2010. All procedures were performed under spinal anesthesia. Postprocedure BT planning CT scans were obtained with patients in supine position with arms on the chest (interslice thickness of 3 mm). Thereafter, verification CT was repeated at every alternate fraction. Needle displacements were measured in reference to a relocatable bony point. The mean cranial, caudal, anteroposterior, and mediolateral displacements were recorded. Statistical significance of mean interfraction displacements was evaluated with Wilcoxon Test. RESULTS: Twenty patients were included. Seventeen received boost BT (20 Gy/5 fractions/3 days) after external radiation, three received radical BT alone (36 Gy/9 fractions/5-8 days). An average of three scans (range, 2-3) were available per patient, and 357 needle displacements were analyzed. For the entire study cohort, the average of mean needle displacement was 2.5 mm (range, 0-7.4), 17.4 mm (range, 0-27.9), 1.7 mm (range, 0-6.7), 2.1 mm (range, 0-9.5), 1.7 mm (range, 0-9.3), and 0.6 mm (range, 0-7.8) in cranial, caudal, anterior, posterior, right, and left directions, respectively. The mean displacement in the caudal direction was higher between Days 1 and 2 than that between Days 2 and 3 (13.4 mm vs. 3.8 mm; p = 0.01). The average caudal displacements were no different between reirradiation and boost cohort (15.2 vs. 17.8 mm). CONCLUSIONS: Clinically significant caudal displacements occur during multifractionated pelvic brachytherapy. Optimal margins need to be incorporated while preplanning brachytherapy to account for interfraction displacements.