Class solution for inversely planned permanent prostate implants to mimic an experienced dosimetrist

Conventional vs machine learning-based treatment planning in prostate brachytherapy: Results of a Phase I randomized controlled trial

PURPOSE

The purpose of this study was to evaluate the noninferiority of Day 30 dosimetry between a machine learning-based treatment planning system for prostate low-dose-rate (LDR) brachytherapy and the conventional, manual planning technique. As a secondary objective, the impact of planning technique on clinical workflow efficiency was also evaluated.

MATERIALS AND METHODS

41 consecutive patients who underwent I-125 LDR monotherapy for low- and intermediate-risk prostate cancer were accrued into this single-institution study between 2017 and 2018. Patients were 1:1 randomized to receive treatment planning using a machine learning-based prostate implant planning algorithm (PIPA system) or conventional, manual technique. Treatment plan modifications by the radiation oncologist were evaluated by computing the Dice coefficient of the prostate V_150% isodose volume between either the PIPA-or conventional-and final approved plans. Additional evaluations between groups evaluated the total planning time and dosimetric outcomes at preimplant and Day 30.

RESULTS

21 and 20 patients were treated using the PIPA and conventional techniques, respectively. No significant differences were observed in preimplant or Day 30 prostate D_90%, V_100%, rectum V₁₀₀, or rectum D_1cc between PIPA and conventional techniques. Although the PIPA group had a larger proportion of patients with plans requiring no modifications (Dice = 1.00), there was no significant difference between the magnitude of modifications between each arm. There was a large significant advantage in mean planning time for the PIPA arm (2.38 ± 0.96 min) compared with the conventional (43.13 ± 58.70 min) technique (p >> 0.05).

CONCLUSIONS

A machine learning-based planning workflow for prostate LDR brachytherapy has the potential to offer significant time savings and operational efficiencies, while producing noninferior postoperative dosimetry to that of expert, conventional treatment planners.

Evaluation of a Machine-Learning Algorithm for Treatment Planning in Prostate Low-Dose-Rate Brachytherapy

PURPOSE

This work presents the application of a machine learning (ML) algorithm to automatically generate high-quality, prostate low-dose-rate (LDR) brachytherapy treatment plans. The ML algorithm can mimic characteristics of preoperative treatment plans deemed clinically acceptable by brachytherapists. The planning efficiency, dosimetry, and quality (as assessed by experts) of preoperative plans generated with an ML planning approach was retrospectively evaluated in this study.

METHODS AND MATERIALS

Preimplantation and postimplantation treatment plans were extracted from 100 high-quality LDR treatments and stored within a training database. The ML training algorithm matches similar features from a new LDR case to those within the training database to rapidly obtain an initial seed distribution; plans were then further fine-tuned using stochastic optimization. Preimplantation treatment plans generated by the ML algorithm were compared with brachytherapist (BT) treatment plans in terms of planning time (Wilcoxon rank sum, α = 0.05) and dosimetry (1-way analysis of variance, α = 0.05). Qualitative preimplantation plan quality was evaluated by expert LDR radiation oncologists using a Likert scale questionnaire.

RESULTS

The average planning time for the ML approach was 0.84 ± 0.57 minutes, compared with 17.88 ± 8.76 minutes for the expert planner (P=.020). Preimplantation plans were dosimetrically equivalent to the BT plans; the average prostate V150% was 4% lower for ML plans (P=.002), although the difference was not clinically significant. Respondents ranked the ML-generated plans as equivalent to expert BT treatment plans in terms of target coverage, normal tissue avoidance, implant confidence, and the need for plan modifications. Respondents had difficulty differentiating between plans generated by a human or those generated by the ML algorithm.

CONCLUSIONS

Prostate LDR preimplantation treatment plans that have equivalent quality to plans created by brachytherapists can be rapidly generated using ML. The adoption of ML in the brachytherapy workflow is expected to improve LDR treatment plan uniformity while reducing planning time and resources.

An open-source genetic algorithm for determining optimal seed distributions for low-dose-rate prostate brachytherapy

PURPOSE

An open source optimizer that generates seed distributions for low-dose-rate prostate brachytherapy was designed, tested, and validated.

METHODS

The optimizer was a simple genetic algorithm (SGA) that, given a set of prostate and urethra contours, determines the optimal seed distribution in terms of coverage of the prostate with the prescribed dose while avoiding hotspots within the urethra. The algorithm was validated in a retrospective study on 45 previously contoured low-dose-rate prostate brachytherapy patients. Dosimetric indices were evaluated to ensure solutions adhered to clinical standards. The SGA performance was further benchmarked by comparing solutions obtained from a commercial optimizer (inverse planning simulated annealing [IPSA]) with the same cohort of 45 patients.

RESULTS

Clinically acceptable target coverage by the prescribed dose (V100) was obtained for both SGA and IPSA, with a mean ± standard deviation of 98 ± 2% and 99.5 ± 0.5%, respectively. For the prostate D90, SGA and IPSA yielded 177 ± 8 Gy and 186 ± 7 Gy, respectively, which were both clinically acceptable. Both algorithms yielded reasonable dose to the rectum, with V100 < 0.3 cc. A reduction in dose to the urethra was seen using SGA. SGA solutions showed a slight prostate volume dependence, with smaller prostates (<25 cc) yielding less desirable, although still clinically viable, dosimetric outcomes. SGA plans used, on average, fewer needles than IPSA (21 vs. 24, respectively), which may lead to a reduction in urinary toxicity and edema that alters post-implant dosimetry.

CONCLUSIONS

An open source SGA was validated that provides a research tool for the brachytherapy community.

Does help structures play a role in reducing the variation of dwell time in IPSA planning for gynaecological brachytherapy application?

PURPOSE

To report our experience of dosimetric comparison of IPSA and manual plans, with a focus on the use of help structures (HS) during optimization.

MATERIAL AND METHODS

33 patients who underwent MR image-based HDR intracavitary-brachytherapy for cervix cancer based on GYN-ESTRO recommendations were selected for evaluation. Tandem/ovoid (T/O) and Vienna applicators were used. HS of diameter of 5 mm were drawn around the tandem/needles/ovoid and ring. Three plans were generated: manual optimized plan (MOPT), IPSA without help structures (IPSA_woHS) and IPSA with help structures (IPSA_wHS). Dose-volume parameters and the loading pattern were evaluated.

RESULTS

For T/O, the use of HS did not make significant impact in the dose-volume parameters and in the loading of tandem and ovoids, however steep variation was found in the individual dwell time. In case of Vienna applicator, inclusion of HS in the optimization made a significant impact in loading of needles. The percentage ratio of total time of needles to the tandem (T(N/T%)) was found to be 14±2.5, 53±9, 22±6 for MOPT, IPSA_woHS and IPSA_wHS, respectively, which implies that in IPSA_woHS the dwell time in needles were half of the dwell time in the tandem, while in MOPT the needles were loaded only in 14%, and in IPSA_wHS it was 22% of the dwell time of tandem. Inclusion of HS in the optimization has reduced the contribution of dwell time of needle in IPSA_wHS. The individual variation of dwell time was also reduced in IPSA_wHS, however drawing of HS is a time consuming procedure and may not be practical for a routine practice.

CONCLUSION

The role of HS was evaluated for IPSA for T/O and Vienna-applicator, the use of HS may be beneficial in case of combined intracavitary - interstitial approach.

A detailed dosimetric comparison between manual and inverse plans in HDR intracavitary/interstitial cervical cancer brachytherapy

PURPOSE

The purpose of this study was to compare two inverse planning algorithms for cervical cancer brachytherapy and a conventional manual treatment planning according to the MUW (Medical University of Vienna) protocol.

MATERIAL AND METHODS

For 20 patients, manually optimized, and, inversely optimized treatment plans with Hybrid Inverse treatment Planning and Optimization (HIPO) and with Inverse Planning Simulated Annealing (IPSA) were created. Dosimetric parameters, absolute volumes of normal tissue receiving reference doses, absolute loading times of tandem, ring and interstitial needles, Paddick and COIN conformity indices were evaluated.

RESULTS

HIPO was able to achieve a similar dose distribution to manual planning with the restriction of high dose regions. It reduced the loading time of needles and the overall treatment time. The values of both conformity indices were the lowest. IPSA was able to achieve acceptable dosimetric results. However, it overloaded the needles. This resulted in high dose regions located in the normal tissue. The Paddick index for the volume of two times prescribed dose was outstandingly low.

CONCLUSIONS

HIPO can produce clinically acceptable treatment plans with the elimination of high dose regions in normal tissue. Compared to IPSA, it is an inverse optimization method which takes into account current clinical experience gained from manual treatment planning.

Toward adaptive stereotactic robotic brachytherapy for prostate cancer: demonstration of an adaptive workflow incorporating inverse planning and an MR stealth robot

To translate any robot into a clinical environment, it is critical that the robot can seamlessly integrate with all the technology of a modern clinic. MRBot, an MR-stealth brachytherapy delivery device, was used in a closed-bore 3T MRI and a clinical brachytherapy cone beam CT suite. Targets included ceramic dummy seeds, MR-Spectroscopy-sensitive metabolite, and a prostate phantom. Acquired DICOM images were exported to planning software to register the robot coordinates in the imager's frame, contour and verify target locations, create dose plans, and export needle and seed positions to the robot. The coordination of each system element (imaging device, brachytherapy planning system, robot control, robot) was validated with a seed delivery accuracy of within 2 mm in both a phantom and soft tissue. An adaptive workflow was demonstrated by acquiring images after needle insertion and prior to seed deposition. This allows for adjustment if the needle is in the wrong position. Inverse planning (IPSA) was used to generate a seed placement plan and coordinates for ten needles and 29 seeds were transferred to the robot. After every two needles placed, an image was acquired. The placed seeds were identified and validated prior to placing the seeds in the next two needles. The ability to robotically deliver seeds to locations determined by IPSA and the ability of the system to incorporate novel needle patterns were demonstrated. Shown here is the ability to overcome this critical step. An adaptive brachytherapy workflow is demonstrated which integrates a clinical anatomy-based seed location optimization engine and a robotic brachytherapy device. Demonstration of this workflow is a key element of a successful translation to the clinic of the MRI stealth robotic delivery system, MRBot.

Inverse planning optimization for hybrid prostate permanent-seed implant brachytherapy plans using two source strengths

The purpose is to demonstrate the ability to generate clinically acceptable prostate permanent seed implant plans using two seed types which are identical except for their activity. The IPSA inverse planning algorithms were modified to include multiple dose matrices for the calculation of dose from different sources, and a selection algorithm was implemented to allow for the swapping of source type at any given source position. Five previously treated patients with a range of prostate volumes from 20-48 cm3 were re-optimized under two hybrid scenarios: (1) using 0.32 and 0.51 mGy m2 / h 125I, and (2) using 0.64 and 0.76 mGy m2 / h 125I. Isodose lines were generated and dosimetric indices , V150Prostate, D90Prostate, V150Urethra, V125Urethra, V120Urethra,V100Urethra, and D10Urethra were calculated. The algorithm allows for the generation of single-isotope, multi-activity hybrid brachytherapy plans. By dealing with only one radionuclide, but of different activity, the biology is unchanged from a standard plan. All V100Prostate were within 2.3 percentage points for every plan and always above the clinically desirable 95%. All V150Urethra were identically zero, and V120Urethra is always below the clinically acceptable value of 1.0 cm3. Clinical optimization times for the hybrid plans are still under one minute, for most cases. It is possible to generate clinically advantageous brachytherapy plans (i.e. obtain the same quality dose distribution as a standard single-activity plan) while incorporating leftover seeds from a previous patient treatment. This method will allow a clinic to continue to provide excellent patient care, but at a reduced cost. Multi-activity hybrid plans were equal in quality (as measured by the standard dosimetric indices) to plans with seeds of a single activity. Despite the expanded search space, optimization times for these studies were still under two minutes on a modern day laptop and can be reduced to below one minute in a clinical setting. With the typical cost of a set of PPI seeds on the order of thousands of dollars, it is possible to reduce the cost of brachytherapy treatments by allowing for easier use of seeds left over from a previous patient or unused due to a cancelled treatment.

Model simulation and experimental validation of intratumoral chemotherapy using multiple polymer implants

Radiofrequency ablation has emerged as a minimally invasive option for liver cancer treatment, but local tumor recurrence is common. To eliminate residual tumor cells in the ablated tumor, biodegradable polymer millirods have been designed for local drug (e.g., doxorubicin) delivery. A limitation of this method has been the extent of drug penetration into the tumor (<5 mm), especially in the peripheral tumor rim where thermal ablation is less effective. To provide drug concentration above the therapeutic level as needed throughout a large tumor, implant strategies with multiple millirods were devised using a computational model. This dynamic, 3-D mass balance model of drug distribution in tissue was used to simulate the consequences of various numbers of implants in different locations. Experimental testing of model predictions was performed in a rabbit VX2 carcinoma model. This study demonstrates the value of multiple implants to provide therapeutic drug levels in large ablated tumors.