Differential Diagnosis of Prostate Cancer Grade to Augment Clinical Diagnosis Based on Classifier Models with Tuned Hyperparameters.

We developed a novel machine-learning algorithm to augment the clinical diagnosis of prostate cancer utilizing first and second-order texture analysis metrics in a novel application of machine-learning radiomics analysis. We successfully discriminated between significant prostate cancers versus non-tumor regions and provided accurate prediction between Gleason score cohorts with statistical sensitivity of 0.82, 0.81 and 0.91 in three separate pathology classifications. Tumor heterogeneity and prediction of the Gleason score were quantified using two feature selection approaches and two separate classifiers with tuned hyperparameters. There was a total of 71 patients analyzed in this study. Multiparametric MRI, incorporating T2WI and ADC maps, were used to derive radiomics features. Recursive feature elimination (RFE), the least absolute shrinkage and selection operator (LASSO), and two classification approaches, incorporating a support vector machine (SVM) (with randomized search) and random forest (RF) (with grid search), were utilized to differentiate between non-tumor regions and significant cancer while also predicting the Gleason score. In T2WI images, the RFE feature selection approach combined with RF and SVM classifiers outperformed LASSO with SVM and RF classifiers. The best performance was achieved by combining LASSO and SVM into a model that used both T2WI and ADC images. This model had an area under the curve (AUC) of 0.91. Radiomic features computed from ADC and T2WI images were used to predict three groups of Gleason score using two kinds of feature selection methods (RFE and LASSO), RF and SVM classifier models with tuned hyperparameters. Using combined sequences (T2WI and ADC map images) and combined radiomics (1st and GLCM features), LASSO, with a feature selection method with RF, was able to predict G3 with the highest sensitivity at a level AUC of 0.92. To predict G3 for single sequence (T2WI images) using GLCM features, LASSO with SVM achieved the highest sensitivity with an AUC of 0.92.

Systematic evaluation of spatial resolution and gamma criteria for quality assurance with detector arrays in stereotactic radiosurgery.

PURPOSE: To characterize detector array spacing and gamma index for quality assurance (QA) of stereotactic radiosurgery (SRS) deliveries. Use the Nyquist theorem to determine the required detector spacing in SRS fields, and find optimal gamma indices to detect MLC errors using the SRS MapCHECK, ArcCHECK, and a portal imaging device (EPID). METHODS: The required detector spacing was determined via Fourier analysis of small radiation fields and profiles of typical SRS treatment plans. The clinical impact of MLC errors of 0.5, 1, and 2 mm was evaluated. Global gamma (low-dose threshold 10%) was evaluated for the three detector systems using various combinations of the distance to agreement and the dose difference. RESULTS: While MLC errors only slightly affected mean dose to PTV and a 2 mm thick surrounding structure (PTV_2 mm), significant PTV underdose incurred with increase in maximum dose to PTV_2 mm. Gamma indices with highest sensitivity to the introduced errors at 95% tolerance level for plans on target volumes of 3.2 cm3 (plan 3 cc) and 35.02 cm3 (plan 35 cc) were 2%/1 mm for the SRS MapCHECK and 2%/3 mm for the ArcCHECK, with 3%/1 mm (plan 3cc) and 2%/1 mm (plan 35cc) for the EPID. Drops in passing rates for a 2 mm MLC error were (46.2%, 41.6%) for the SRS MapCHECK and (12.2%, 4.2%) for the ArcCHECK for plan 3cc and plan 35cc, respectively. For Portal Dose, values were 4.5% (plan 3cc) and 7% (plan 35cc). The Nyquist frequency of two SRS dose distributions lie between 0.26  and 0.1 mm-1 , corresponding to detector spacings of 1.9 and 5 mm. Evaluation of SRS MapCHECK data with doubled detector density indicates that increased detector density may reduce the system's sensitivity to errors, necessitating a tighter gamma index. CONCLUSIONS: The present results give insight on the performance of detector arrays and gamma indices for the investigated detectors during SRS QA.

Whole-brain traumatic controlled cortical impact to the left frontal lobe: Magnetic resonance image-based texture analysis.

This research assesses the capability of texture analysis (TA) derived from high-resolution (HR) T2-weighted magnetic resonance imaging to identify primary sequelae following 1-5 hours of controlled cortical impact mild or severe traumatic brain injury (TBI) to the left frontal cortex (focal impact) and secondary (diffuse) sequelae in the right frontal cortex, bilateral corpus callosum, and hippocampus in rats. The TA technique comprised first-order (histogram-based) and second-order statistics (including gray-level co-occurrence matrix, gray-level run length matrix, and neighborhood gray-level difference matrix). Edema in the left frontal impact region developed within 1 hour and continued throughout the 5-hour assessments. The TA features from HR images confirmed the focal injury. There was no significant difference among radiomics features between the left and right corpus callosum or hippocampus from 1 to 5 hours following a mild or severe impact. The adjacent corpus callosum region and the distal hippocampus region (s), showed no diffuse injury 1-5 hours after mild or severe TBI. These results suggest that combining HR images with TA may enhance detection of early primary and secondary sequelae following TBI.

Analysis of CyberKnife intracranial treatment plans using ICRU 91 dose reporting: A retrospective study.

ICRU 91, published in 2017, is an international standard for prescribing, recording, and reporting stereotactic treatments. Since its release, there has been limited research published on the implementation and impact of ICRU 91 on clinical practice. This work provides an assessment of the recommended ICRU 91 dose reporting metrics for their use in clinical treatment planning. A set of 180 intracranial stereotactic treatment plans for patients treated by the CyberKnife (CK) system were analyzed retrospectively using the ICRU 91 reporting metrics. The 180 plans comprised 60 trigeminal neuralgia (TGN), 60 meningioma (MEN), and 60 acoustic neuroma (AN) cases. The reporting metrics included the planning target volume (PTV) near-minimum dose ( D near - min ${D}_{{\rm{near}} - {\rm{min}}}$ ), near-maximum dose ( D near - max ${D}_{{\rm{near}} - {\rm{max}}}$ ), and median dose ( D 50 % ${D}_{50{\rm{\% }}}$ ), as well as the gradient index (GI) and conformity index (CI). The metrics were assessed for statistical correlation with several treatment plan parameters. In the TGN plan group, owing to the small targets, D near - min ${D}_{{\rm{near}} - {\rm{min}}}$ was greater than D near - max ${D}_{{\rm{near}} - {\rm{max}}}$ in 42 plans, whereas both metrics were not applicable in 17 plans. The D 50 % ${D}_{50{\rm{\% }}}$ metric was predominantly influenced by the prescription isodose line (PIDL). The GI was significantly dependent on target volume in all analyses performed, where the variables were inversely related. The CI was only dependent on target volume in treatment plans for small targets. The ICRU 91 D near - min ${D}_{{\rm{near}} - {\rm{min}}}$ and D near - max ${D}_{{\rm{near}} - {\rm{max}}}$ metrics breakdown in plans for small target volumes below 1 cm3 ; the Min and Max pixel should be reported in such cases. The D 50 % ${D}_{50{\rm{\% }}}$ metric is of limited use for treatment planning. Given their volume dependence, the GI and CI metrics could potentially serve as plan evaluation tools in the planning of the sites analyzed in this study, which would ultimately improve treatment plan quality.

Compensation for radiotherapy treatment interruptions due to a cyberattack: An isoeffective DVH-based dose compensation decision tool.

Unscheduled interruptions to radiotherapy treatments lead to decreased tumor control probability (TCP). Rapid cell repopulation in the tumor increases due to the absence of radiation dose, resulting in the loss of TCP. Compensation for this loss is required to prevent or reduce an extension of the patient's overall treatment time and regain the original TCP. The cyberattack on the Irish public health service in May 2021 prevented radiotherapy treatment delivery resulting in treatment interruptions of up to 12 days. Current standards for treatment gap calculations are performed using the Royal College of Radiologists (RCR) methodology, using a point-dose for planning target volume (PTV) and the organs at risk (OAR). An in-house tool, named EQD2 VH, was created in Python to perform treatment gap calculations using the dose-volume histogram (DVH) information in DICOM data extracted from commercial treatment planning system plans. The physical dose in each dose bin was converted into equivalent dose in 2-Gy fractions (EQD2 ), accounting for tumor cell repopulation. This EQD2 -based DVH provides a 2D representation of the impact of treatment gap compensation strategies on both PTV and OAR dose distributions compared to the intended prescribed treatment plan. This additional information can aid clinicians' choice of compensation options. EQD2 VH was evaluated using five high-priority patients experiencing a treatment interruption when the cyberattack occurred. Compensation plans were created using the RCR methodology to evaluate EQD2 VH as a decision-making tool. The EQD2 VH method demonstrated that the comparison of compensated treatment plans alongside the original intended treatment plans using isoeffective DVH analysis can be achieved. It enabled a visual and quantitative comparison between treatment plan options and provided an individual analysis of each structure in a patient's plan. It demonstrated potential to be a useful decision-making tool for finding a balance between optimizing dose to PTV while protecting OARs.

Quantifying Tumor Heterogeneity from Multiparametric Magnetic Resonance Imaging of Prostate Using Texture Analysis.

(1) Background: Multiparametric MRI (mp-MRI) is used to manage patients with PCa. Tumor identification via irregular sampling or biopsy is problematic and does not allow the comprehensive detection of the phenotypic and genetic alterations in a tumor. A non-invasive technique to clinically assess tumor heterogeneity is also in demand. We aimed to identify tumor heterogeneity from multiparametric magnetic resonance images using texture analysis (TA). (2) Methods: Eighteen patients with prostate cancer underwent mp-MRI scans before prostatectomy. A single radiologist matched the histopathology report to single axial slices that best depicted tumor and non-tumor regions to generate regions of interest (ROIs). First-order statistics based on the histogram analysis, including skewness, kurtosis, and entropy, were used to quantify tumor heterogeneity. We compared non-tumor regions with significant tumors, employing the two-tailed Mann-Whitney U test. Analysis of the area under the receiver operating characteristic curve (ROC-AUC) was used to determine diagnostic accuracy. (3) Results: ADC skewness for a 6 × 6 px filter was significantly lower with an ROC-AUC of 0.82 (p = 0.001). The skewness of the ADC for a 9 × 9 px filter had the second-highest result, with an ROC-AUC of 0.66; however, this was not statistically significant (p = 0.08). Furthermore, there were no substantial distinctions between pixel filter size groups from the histogram analysis, including entropy and kurtosis. (4) Conclusions: For all filter sizes, there was poor performance in terms of entropy and kurtosis histogram analyses for cancer diagnosis. Significant prostate cancer may be distinguished using a textural feature derived from ADC skewness with a 6 × 6 px filter size.

A comparison of treatment planning techniques for low-dose-rate (LDR) prostate brachytherapy.

PURPOSE: The purpose of this study was to compare low-dose-rate prostate brachytherapy treatment plans created using three retrospectively applied planning techniques with plans delivered to patients. METHODS AND MATERIALS: Treatment plans were created retrospectively on transrectal ultrasound (TRUS) scans for 26 patients. The technique dubbed 4D Brachytherapy was applied, using TRUS and MRI to obtain prostatic measurements required for the associated webBXT online nomogram. Using a patient's MRI scan to create a treatment plan involving loose seeds was also explored. Plans delivered to patients were made using an intraoperative loose seed TRUS-based planning technique. Prostate V100 (%), prostate V150 (%), prostate D90 (Gy), rectum D0.1cc (Gy), rectum D2cc (Gy), urethra D10 (%), urethra D30 (%), and prostate volumes were measured for each patient. Statistical analysis was used to assess and compare plans. RESULTS: Prostate volumes measured by TRUS and MRI were significantly different. Prostate volumes calculated by the webBXT online nomogram using TRUS- and MRI-based measurements were not significantly different. Compared with delivered plans, TRUS-based 4D Brachytherapy plans showed significantly lower rectum D0.1cc (Gy) values, MRI-based 4D Brachytherapy plans showed significantly higher prostate V100 (%) values and significantly lower rectum D0.1cc (Gy), urethra D10 (%), and urethra D30 (%) values, and loose seed MRI-based plans showed significantly lower prostate V100 (%), prostate D90 (Gy), rectum D0.1cc (Gy), rectum D2cc (Gy), urethra D10 (%), and urethra D30 (%) values. CONCLUSIONS: TRUS-based 4D Brachytherapy plans showed similar dosimetry to delivered plans; rectal dosimetry was superior. MRI can be integrated into the 4D Brachytherapy workflow. The webBXT online nomogram overestimates the required number of seeds.

Off-axis dose distribution with stand-in and stand-off configurations for superficial radiotherapy treatments.

Current practice when delivering dose for superficial skin radiotherapy is to adjust the monitor units so that the prescribed dose is delivered to the central axis of the superficial unit applicator. Variations of source-to-surface distance due to patient's anatomy protruding into the applicator or extending away from the applicator require adjustments to the monitor units using the inverse square law. Off-axis dose distribution varies significantly from the central axis dose and is not currently being quantified. The dose falloff at the periphery of the field is not symmetrical in the anode-cathode axis due to the heel effect. This study was conducted to quantify the variation of dose across the surface being treated and model a simple geometric shape to estimate a patient's surface with stand-in and stand-off. Isodose plots and color-coded dose distribution maps were produced from scans of GAFChromic EBT-3 film irradiated by a Gulmay D3300 orthovoltage x-ray therapy system. It was clear that larger applicators show a greater dose falloff toward the periphery than smaller applicators. Larger applicators were found to have a lower percentage of points above 90% of central axis dose (SA90). Current clinical practice does not take this field variation into account. Stand-in can result in significant dose falloff off-axis depending on the depth and width of the protrusion, while stand-off can result in a flatter field due to the high-dose region near the central axis being further from the source than the peripheral regions. The central axis also received a 7% increased or decreased dose for stand-in or stand-off, respectively.

Optimal tumour control for early-stage non-small-cell lung cancer: A radiobiological modelling perspective.

A fully heterogeneous population tumour control probability (TCP) model, based on the linear-quadratic (LQ) cell survival concept combined with the Poisson statistic, was established to predict local tumour control after one, two and three years. This TCP model was created using data from 16 publications that reported on early-stage non-small-cell lung cancer (NSCLC) treated using either three-dimensional conformal radiation therapy (3D-CRT), continuous hyperfractionated accelerated radiotherapy (CHART) or stereotactic ablative body radiotherapy (SABR). The TCP model was fitted to the clinical outcome data using optimised radiosensitivity values produced by the Nelder-Mead simplex algorithm. The statistical analysis resulted in R2 values of 0.96, 0.96 and 0.97 and wRMSE values of 3.9%, 5.2% and 5.9% for one-, two- and three-year local tumour control rates, respectively. The TCP models for one, two and three years were internally validated using a bootstrap resampling approach. The mean R2 and 95% CI for the bootstrap samples were 0.98 (0.93-0.99), 0.98 (0.95-0.99) and 0.98 (0.96-0.99) for the one-, two- and three-year local tumour control rates, respectively. Variations in the TCP with clonogenic density were then further investigated by introducing a new mathematical model to vary the clonogenic cell and radiation dose distribution across the treated volume. Based on the above model, it was estimated that 60% of the dose was sufficient to maintain the TCP after two years for the areas with lower clonogenic cell density. If externally validated, this lower-dose treatment plan could have beneficial effects on the surrounding healthy tissue without negatively affecting tumour control.