Publications by authors named "Jason W Chan"

20 Publications

  • Page 1 of 1

Attention-Aware Discrimination for MR-to-CT Image Translation Using Cycle-Consistent Generative Adversarial Networks.

Radiol Artif Intell 2020 Mar 25;2(2):e190027. Epub 2020 Mar 25.

Department of Radiation Oncology, University of California, 1600 Divisidero St, San Francisco, CA 94115.

Purpose: To suggest an attention-aware, cycle-consistent generative adversarial network (A-CycleGAN) enhanced with variational autoencoding (VAE) as a superior alternative to current state-of-the-art MR-to-CT image translation methods.

Materials And Methods: An attention-gating mechanism is incorporated into a discriminator network to encourage a more parsimonious use of network parameters, whereas VAE enhancement enables deeper discrimination architectures without inhibiting model convergence. Findings from 60 patients with head, neck, and brain cancer were used to train and validate A-CycleGAN, and findings from 30 patients were used for the holdout test set and were used to report final evaluation metric results using mean absolute error (MAE) and peak signal-to-noise ratio (PSNR).

Results: A-CycleGAN achieved superior results compared with U-Net, a generative adversarial network (GAN), and a cycle-consistent GAN. The A-CycleGAN averages, 95% confidence intervals (CIs), and Wilcoxon signed-rank two-sided test statistics are shown for MAE (19.61 [95% CI: 18.83, 20.39], = .0104), structure similarity index metric (0.778 [95% CI: 0.758, 0.798], = .0495), and PSNR (62.35 [95% CI: 61.80, 62.90], = .0571).

Conclusion: A-CycleGANs were a superior alternative to state-of-the-art MR-to-CT image translation methods.© RSNA, 2020.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1148/ryai.2020190027DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC8017410PMC
March 2020

Timing of Urgent Inpatient Palliative Radiation Therapy.

Adv Radiat Oncol 2021 May-Jun;6(3):100670. Epub 2021 Feb 11.

Department of Radiation Oncology, University of California, San Francisco, California.

Purpose: Urgent indications for palliative radiation therapy (RT) include malignant spinal cord compression, symptomatic brain metastases, pain, airway obstruction, and bleeding. Data on the timing of palliative RT in the inpatient setting are limited. We report our experience with inpatient palliative RT at a tertiary academic center and evaluate the effect of a dedicated inpatient palliative RT nurse practitioner (NP) on treatment timelines.

Methods And Materials: We performed a retrospective, single-institution review of 219 inpatients consulted for RT to sites of metastatic disease between May 2012 and May 2018. We compared time-to-treatment intervals before and after integrating an NP for palliative RT in August 2017.

Results: The median age of the 219 patients receiving RT was 61 years (interquartile range [IQR], 51-69 years). The most frequent indications were symptomatic brain metastases (73 patients [33%]), pain (61 patients [28%]), and cord/cauda compression (48 patients [22%]). The median time from consultation request to consult was 1 day (IQR, 0-2 days), and the median time from consultation request to first RT fraction was 3 days (IQR, 2-6 days). The median time from consultation request to RT was shorter for cord compression (2 [IQR, 1-4] days) than for pain (5 [IQR, 2-7] days) ( = .001) or symptomatic brain metastases (3 [IQR, 1-6] days; = .037). With an NP, patients were more likely to undergo same-day consultation and simulation (75% vs 60%; = .045), which was associated with shorter median duration from consultation to initiation of RT (1 [IQR, 0-3] days vs 4 [IQR, 2-7] days; <.001). After the integration of an NP for palliative RT, patients had a higher median Karnofsky Performance Score (70 [IQR, 60-80] vs 50 [IQR, 40-60]; < .001) and were more likely to complete their prescribed RT course (93% vs 82%; = .05).

Conclusions: Time from consultation request to RT is necessarily short for urgent inpatient palliative RT. Advanced practice providers may facilitate and potentially expedite treatment, with significantly shorter times to treatment among patients who undergo same-day consultation and simulation.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.adro.2021.100670DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC8005735PMC
February 2021

Reduced-Dose Radiation Therapy for HPV-Associated Oropharyngeal Carcinoma (NRG Oncology HN002).

J Clin Oncol 2021 Mar 28;39(9):956-965. Epub 2021 Jan 28.

Stanford University, Stanford, CA.

Purpose: Reducing radiation treatment dose could improve the quality of life (QOL) of patients with good-risk human papillomavirus-associated oropharyngeal squamous cell carcinoma (OPSCC). Whether reduced-dose radiation produces disease control and QOL equivalent to standard chemoradiation is not proven.

Patients And Methods: In this randomized, phase II trial, patients with p16-positive, T1-T2 N1-N2b M0, or T3 N0-N2b M0 OPSCC (7th edition staging) with ≤ 10 pack-years of smoking received 60 Gy of intensity-modulated radiation therapy (IMRT) over 6 weeks with concurrent weekly cisplatin (C) or 60 Gy IMRT over 5 weeks. To be considered for a phase III study, an arm had to achieve a 2-year progression-free survival (PFS) rate superior to a historical control rate of 85% and a 1-year mean composite score ≥ 60 on the MD Anderson Dysphagia Inventory (MDADI).

Results: Three hundred six patients were randomly assigned and eligible. Two-year PFS for IMRT + C was 90.5% rejecting the null hypothesis of 2-year PFS ≤ 85% ( = .04). For IMRT, 2-year PFS was 87.6% ( = .23). One-year MDADI mean scores were 85.30 and 81.76 for IMRT + C and IMRT, respectively. Two-year overall survival rates were 96.7% for IMRT + C and 97.3% for IMRT. Acute adverse events (AEs) were defined as those occurring within 180 days from the end of treatment. There were more grade 3-4 acute AEs for IMRT + C (79.6% 52.4%; < .001). Rates of grade 3-4 late AEs were 21.3% and 18.1% ( = .56).

Conclusion: The IMRT + C arm met both prespecified end points justifying advancement to a phase III study. Higher rates of grade ≥ 3 acute AEs were reported in the IMRT + C arm.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1200/JCO.20.03128DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC8078254PMC
March 2021

Gallium-68 prostate-specific membrane antigen ([Ga]Ga-PSMA-11) PET for imaging of thyroid cancer: a feasibility study.

EJNMMI Res 2020 Oct 22;10(1):128. Epub 2020 Oct 22.

Department of Radiology and Biomedical Imaging, University of California San Francisco, 185 Berry Street, Suite 350, Lobby 6, Box 0946, San Francisco, CA, 94143, USA.

Background: Prostate-specific membrane antigen (PSMA) is expressed in the microvasculature of thyroid cancer. This suggests the potential use of PSMA as a diagnostic agent in patients with aggressive forms of thyroid cancer. The purpose of the current study was to determine the feasibility and utility of [Ga]Ga-PSMA-11 PET/MRI in thyroid cancer patients.

Methods: Eligible patients for this prospective pilot study were adults with a history of pathology-proven thyroid cancer who had abnormal radiotracer uptake on an 2-[F]FDG PET and/or I scintigraphy performed in the 12 months prior to study enrollment. Patients underwent a [Ga]Ga-PSMA-11 PET/MRI, and comparison was made to the prior qualifying 2-[F]FDG PET CT/MRI for lesion location and relative intensity.

Results: Twelve patients underwent [Ga]Ga-PSMA-11 PET/MRI, one of which was excluded from analysis due to debulking surgery prior to the PSMA PET. Of the remaining patients, 7/11 had differentiated disease (3 papillary, 2 follicular, 2 Hurthle cell) and 4/11 had dedifferentiated disease (2 poorly differentiated papillary, 2 anaplastic). Out of 43 lesions, 41 were visually 2-[F]FDG positive (uptake greater than background, detection rate 95.3%) and 28 were PSMA positive (uptake greater than background, detection rate 65.1%). Uptake was heterogeneous between patients, and in some cases within patients. 3/11 patients (1 poorly differentiated papillary, 2 follicular) had PSMA uptake which was greater than FDG uptake. For the remaining 8 patients, 2-[F]FDG uptake was greater than PSMA. Using one eligibility guideline in the prostate cancer literature for PSMA radioligand therapy (RLT), 8/11 could be considered eligible for possible future PSMA RLT. This was not predictable based on thyroid cancer subtype.

Conclusions: [Ga]Ga-PSMA-11 PET demonstrated lower detection rate when compared to 2-[F]FDG PET for thyroid cancer lesion visualization. Thyroid cancer subtype alone may not be sufficient to predict PSMA uptake, and radiotracer uptake may vary between patients and even within patients.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1186/s13550-020-00720-3DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7581659PMC
October 2020

Evaluation of a National Comprehensive Cancer Network Guidelines-Based Decision Support Tool in Patients With Non-Small Cell Lung Cancer: A Nonrandomized Clinical Trial.

JAMA Netw Open 2020 09 1;3(9):e209750. Epub 2020 Sep 1.

Department of Radiation Oncology, University of California, San Francisco.

Importance: The association of guideline-based decision support with the quality of care in patients with non-small cell lung cancer (NSCLC) is not known.

Objective: To evaluate the association of exposure to the National Comprehensive Cancer Center (NCCN) guidelines with guideline-concordant care and patients' decisional conflict.

Design, Setting, And Participants: A nonrandomized clinical trial, conducted at a tertiary care academic institution, enrolled patients from February 23, 2015, to September 28, 2017. Data analysis was conducted from July 19, 2019, to April 22, 2020. A cohort of 76 patients with NSCLC seen at diagnosis or disease progression and a retrospective cohort of 157 patients treated before the trial were included. Adherence to 6 NCCN recommendations were evaluated: (1) smoking cessation counseling, (2) adjuvant chemotherapy for patients with stage IB to IIB NSCLC after surgery, (3) pathologic mediastinal staging in patients with stage III NSCLC before surgery, (4) pathologic mediastinal staging in patients with stage III NSCLC before nonsurgical treatment, (5) definitive chemoradiotherapy for patients with stage III NSCLC not having surgery, and (6) molecular testing for epidermal growth factor receptor and anaplastic lymphoma kinase alterations for patients with stage IV NSCLC. Subgroup analysis was conducted to compare the rates of guideline concordance between the prospective and retrospective cohorts. Secondary end points included decisional conflict and satisfaction.

Interventions: An online tool customizing the NCCN guidelines to patients' clinical and pathologic features was used during consultation, facilitated by a trained coordinator.

Main Outcomes And Measures: Concordance of practice with 6 NCCN treatment recommendations on NSCLC and patients' decisional conflict.

Results: Of the 76 patients with NSCLC, 44 were men (57.9%), median age at diagnosis was 68 years (interquartile range [IQR], 41-87 years), and 59 patients (77.6%) had adenocarcinoma. In the retrospective cohort, 91 of 157 patients (58.0%) were men, median age at diagnosis was 66 years (IQR, 61-65 years), and 105 patients (66.9%) had adenocarcinoma. After the intervention, patients received more smoking cessation counseling (4 of 5 [80.0%] vs 1 of 24 [4.2%], P < .001) and less adjuvant chemotherapy (0 of 7 vs 7 of 11 [63.6%]; P = .012). There was no significant change in mutation testing of non-squamous cell stage IV disease (20 of 20 [100%] vs 48 of 57 [84.2%]; P = .10). There was no significant change in pathologic mediastinal staging or initial chemoradiotherapy for patients with stage III disease. After consultation with the tool, decisional conflict scores improved by a median of 20 points (IQR, 3-34; P < .001).

Conclusions And Relevance: The findings of this study suggest that exposure to the NCCN guidelines is associated with increased guideline-concordant care for 2 of 6 preselected recommendations and improvement in decisional conflict.

Trial Registration: ClinicalTrials.gov Identifier: NCT03982459.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1001/jamanetworkopen.2020.9750DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7527870PMC
September 2020

Will the COVID Pandemic Lead to Uncounted Cancer Deaths in the Future?

Int J Radiat Oncol Biol Phys 2020 10;108(2):351-352

Department of Clinical Oncology, LKS Faculty of Medicine, The University of Hong Kong, Hong Kong, China; Clinical Oncology Center, The University of Hong Kong-Shenzhen Hospital, Shenzhen, China.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.ijrobp.2020.07.003DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7462880PMC
October 2020

DoseGAN: a generative adversarial network for synthetic dose prediction using attention-gated discrimination and generation.

Sci Rep 2020 07 6;10(1):11073. Epub 2020 Jul 6.

Department of Radiation Oncology, University of California, San Francisco, CA, 94115, USA.

Deep learning algorithms have recently been developed that utilize patient anatomy and raw imaging information to predict radiation dose, as a means to increase treatment planning efficiency and improve radiotherapy plan quality. Current state-of-the-art techniques rely on convolutional neural networks (CNNs) that use pixel-to-pixel loss to update network parameters. However, stereotactic body radiotherapy (SBRT) dose is often heterogeneous, making it difficult to model using pixel-level loss. Generative adversarial networks (GANs) utilize adversarial learning that incorporates image-level loss and is better suited to learn from heterogeneous labels. However, GANs are difficult to train and rely on compromised architectures to facilitate convergence. This study suggests an attention-gated generative adversarial network (DoseGAN) to improve learning, increase model complexity, and reduce network redundancy by focusing on relevant anatomy. DoseGAN was compared to alternative state-of-the-art dose prediction algorithms using heterogeneity index, conformity index, and various dosimetric parameters. All algorithms were trained, validated, and tested using 141 prostate SBRT patients. DoseGAN was able to predict more realistic volumetric dosimetry compared to all other algorithms and achieved statistically significant improvement compared to all alternative algorithms for the V and V of the PTV, V of the rectum, and heterogeneity index.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1038/s41598-020-68062-7DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC7338467PMC
July 2020

Improved Tumor Control Related to Radiotherapy Technological Development for Hypopharyngeal Cancer.

Laryngoscope 2021 02 12;131(2):E452-E458. Epub 2020 May 12.

Department of Radiation Oncology, University of California San Francisco, San Francisco, California.

Objectives/hypothesis: Squamous cell carcinoma of the hypopharynx (SCCHP) is associated with worse survival compared to other head and neck subsites. This report quantifies the impact of technological improvements in radiotherapy (RT) on outcomes over 6 decades.

Methods: Patients with SCCHP receiving curative-intent treatment between 1962 and 2015 were retrospectively reviewed. Kaplan-Meier analyses of freedom from local recurrence (FFLR), progression-free survival (PFS), and overall survival (OS) were compared across treatment eras and radiation techniques. Multivariable Cox proportional hazards modeling was performed to specify the effect of RT technique.

Results: One hundred thirty-four patients had a median follow-up of 17 months (IQR = 9-38). There were no differences in staging or use of surgery over time, but use of chemotherapy concurrent with RT increased (P < .001) beginning in the 2000s. The 24-month FFLR using two-dimensional RT (2D-RT), three-dimensional conformal RT (3D-CRT), and intensity-modulated RT (IMRT) was 52%, 55%, and 80%, respectively; 24-month PFS was 39%, 46%, and 73%, respectively; and 24-month OS was 27%, 40%, and 68%, respectively. OS (P = .01), PFS (P = .03), and FFLR (P = 0.02) were improved with IMRT over 2D-RT, and FFLR appeared to be improved over 3D-CRT (P = .06). Controlling for chemotherapy use and other major variables, IMRT produced a strong influence over FFLR (adjusted hazard ratio [HR] = 0.2, 95% confidence interval [CI]: 0.0-1.2, P = .08) and a benefit in OS (adjusted HR = 0.1, 95% CI: 0.0-0.4, P = .005).

Conclusions: Across 6 decades, patient and tumor characteristics remained similar whereas use of chemoradiation increased and IMRT was adopted. The introduction of IMRT was associated with improved FFLR, PFS, and OS, and a reduction in acute toxicity as compared to earlier radiation technologies.

Level Of Evidence: 4 Laryngoscope, 131:E452-E458, 2021.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/lary.28726DOI Listing
February 2021

Salvage Strategies for Neck Recurrences Are Individualized.

Authors:
Jason W Chan

Int J Radiat Oncol Biol Phys 2020 04;106(5):903-904

Department of Radiation Oncology, University of California San Francisco, San Francisco, California.

View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.ijrobp.2019.09.018DOI Listing
April 2020

Dose response and architecture in volume staged radiosurgery for large arteriovenous malformations: A multi-institutional study.

Radiother Oncol 2020 03 10;144:180-188. Epub 2019 Dec 10.

University of California - San Francisco School of Medicine, Department of Neurosurgery, United States.

Background: Optimal treatment paradigm for large arteriovenous malformations (AVMs) is controversial. Volume-staged stereotactic radiosurgery (VS-SRS) provides an effective option for these high-risk lesions, but optimizing treatment for these recalcitrant and rare lesions has proven difficult.

Methods: This is a multi-centered retrospective review of patients treated with a planned prospective volume staging approach to stereotactically treat the entire nidus of an AVM with volume stages separated by intervals of 3-6 months. A total of 9 radiosurgical centers treated 257 patients with VS-SRS between 1991 and 2016. We evaluated near complete response (nCR), obliteration, cure, and overall survival.

Results: With a median age of 33 years old at the time of first SRS volume stage, patients received 2-4 total volume stages and a median follow up of 5.7 years after VS-SRS. The median total AVM nidus volume was 23.25 cc (range: 7.7-94.4 cc) with a median margin dose per stage of 17 Gy (range: 12-20 Gy). Total AVM volume, margin dose per stage, compact nidus, lack of prior embolization, and lack of thalamic location involvement were all associated with improved outcomes. Dose >/= 17.5 Gy was strongly associated with improved rates of nCR, obliteration, and cure. With dose >/= 17.5 Gy, 5- and 10-year cure rates were 33.7% and 76.8% in evaluable patients compared to 23.7% and 34.7% of patients with 17 Gy and 6.4% and 20.6% with <17 Gy per volume-stage (p = 0.004). Obliteration rates in diffuse nidus architecture with <17 Gy were particularly poor with none achieving obliteration compared to 32.3% with doses >/= 17 Gy at 5 years (p = 0.007). Comparatively, lesions with a compact nidus architecture exhibited obliteration rates at 5 years were 10.7% vs 9.3% vs 26.6% for dose >17 Gy vs 17 Gy vs >/=17.5 Gy.

Conclusion: VS-SRS is an option for upfront treatment of large AVMs. Higher dose was associated with improved rates of nCR, obliteration, and cure suggesting that larger volumetric responses may facilitate salvage therapy and optimize the chance for cure.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.radonc.2019.09.019DOI Listing
March 2020

Diagnosis, Staging, Radiation Treatment Response Assessment, and Outcome Prognostication of Head and Neck Cancers Using PET Imaging: A Systematic Review.

PET Clin 2020 Jan;15(1):65-75

Department of Radiation Oncology, Helen Diller Comprehensive Cancer Care Center, University of California San Francisco, 1600 Divisadero Street, Suite H-1031, San Francisco, CA 94143-1708, USA. Electronic address:

This systematic review summarizes the current applications of F-FDG PET imaging in the diagnosis, staging, radiation treatment response assessment, and outcome prognostication of head and neck cancers. For head and neck cancers of unknown primary origin, F-FDG PET/CT increases the likelihood of identifying the primary tumor and establishing the diagnosis. F-FDG PET/CT is important in the accurate staging of locoregionally advanced cases of HNSCC, which can greatly affect recommendations for treatment. Following definitive chemoradiation, F-FDG PET/CT is validated as a means of treatment response assessment. Emerging PET tracers of hypoxia and their potential applications are reviewed.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.cpet.2019.08.010DOI Listing
January 2020

Attention-enabled 3D boosted convolutional neural networks for semantic CT segmentation using deep supervision.

Phys Med Biol 2019 07 2;64(13):135001. Epub 2019 Jul 2.

These two authors contributed equally. Author to whom correspondence should be addressed.

A deeply supervised attention-enabled boosted convolutional neural network (DAB-CNN) is presented as a superior alternative to current state-of-the-art convolutional neural networks (CNNs) for semantic CT segmentation. Spatial attention gates (AGs) were incorporated into a novel 3D cascaded CNN framework to prioritize relevant anatomy and suppress redundancies within the network. Due to the complexity and size of the network, incremental channel boosting was used to decrease memory usage and facilitate model convergence. Deep supervision was used to encourage semantically meaningful deep features and mitigate local minima traps during training. The accuracy of DAB-CNN is compared to seven architectures: a variation of U-Net (UNet), attention-enabled U-Net (A-UNet), boosted U-Net (B-UNet), deeply-supervised U-Net (D-UNet), U-Net with ResNeXt blocks (ResNeXt), life-long learning segmentation CNN (LL-CNN), and deeply supervised attention-enabled U-Net (DA-UNet). The accuracy of each method was assessed based on Dice score compared to manually delineated contours as the gold standard. One hundred and twenty patients who had definitive prostate radiotherapy were used in this study. Training, validation, and testing followed Kaggle competition rules, with 80 patients used for training, 20 patients used for internal validation, and 20 test patients used to report final accuracies. Comparator p -values indicate that DAB-CNN achieved significantly superior Dice scores than all alternative algorithms for the prostate, rectum, and penile bulb. This study demonstrated that attention-enabled boosted convolutional neural networks (CNNs) using deep supervision are capable of achieving superior prediction accuracy compared to current state-of-the-art automatic segmentation methods.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1088/1361-6560/ab2818DOI Listing
July 2019

Dedicated Diagnostic Radiology/Radiation Oncology Rounds: Added Value Beyond Traditional Tumor Boards.

Curr Probl Diagn Radiol 2020 Jul - Aug;49(4):248-253. Epub 2019 May 9.

Department of Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA.

Objectives: We aimed to evaluate the impact of collaborative discussion between diagnostic radiologists and radiation oncologists on radiation oncology management for thoracic oncology patients.

Methods: We reviewed cases presented at multidisciplinary thoracic tumor boards (TTB) (n = 122) and diagnostic radiology/radiation oncology rounds (DR/ROR) (n = 45). Changes in planned radiation management following imaging discussion were categorized-no change, timing change, and treatment volume change. Phase of care was also classified. In DR/ROR, radiation oncologists were surveyed regarding (1) change in radiation oncology management and (2) change in confidence (both 5-point Likert scales).

Results: Discussion of imaging with a radiologist changed radiation oncology management in 31.1% of TTB cases and 68.9% of DR/ROR cases (P < 0.001). Changes to the timing of initiating radiation therapy occurred with similar frequency in the 2 settings (31.1% vs 46.7%, P = 0.063). Changes to target volume occurred more frequently in DR/ROR (35.6% vs <1%), P < 0.001. Over half of imaging discussions in DR/ROR resulted in at least "moderate" change in radiation oncology management, and the level of confidence held by the radiation oncologists increased following discussion with radiologists in 95.6% of cases.

Conclusion: Collaborative discussions between radiation oncologists and diagnostic radiologists in a multispecialty tumor board and in targeted 2-specialty rounds are not redundant, but result in different management changes and at different phases of care. Our study emphasizes the importance of consultation with physicians as an area where radiologists can add value, specifically the added benefit of smaller collaborative discussions.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1067/j.cpradiol.2019.05.004DOI Listing
March 2021

A convolutional neural network algorithm for automatic segmentation of head and neck organs at risk using deep lifelong learning.

Med Phys 2019 May 4;46(5):2204-2213. Epub 2019 Apr 4.

Department of Radiation Oncology, University of California, San Francisco, CA, 94115, USA.

Purpose: This study suggests a lifelong learning-based convolutional neural network (LL-CNN) algorithm as a superior alternative to single-task learning approaches for automatic segmentation of head and neck (OARs) organs at risk.

Methods And Materials: Lifelong learning-based convolutional neural network was trained on twelve head and neck OARs simultaneously using a multitask learning framework. Once the weights of the shared network were established, the final multitask convolutional layer was replaced by a single-task convolutional layer. The single-task transfer learning network was trained on each OAR separately with early stoppage. The accuracy of LL-CNN was assessed based on Dice score and root-mean-square error (RMSE) compared to manually delineated contours set as the gold standard. LL-CNN was compared with 2D-UNet, 3D-UNet, a single-task CNN (ST-CNN), and a pure multitask CNN (MT-CNN). Training, validation, and testing followed Kaggle competition rules, where 160 patients were used for training, 20 were used for internal validation, and 20 in a separate test set were used to report final prediction accuracies.

Results: On average contours generated with LL-CNN had higher Dice coefficients and lower RMSE than 2D-UNet, 3D-Unet, ST- CNN, and MT-CNN. LL-CNN required ~72 hrs to train using a distributed learning framework on 2 Nvidia 1080Ti graphics processing units. LL-CNN required 20 s to predict all 12 OARs, which was approximately as fast as the fastest alternative methods with the exception of MT-CNN.

Conclusions: This study demonstrated that for head and neck organs at risk, LL-CNN achieves a prediction accuracy superior to all alternative algorithms.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/mp.13495DOI Listing
May 2019

Advances in multidisciplinary therapy for meningiomas.

Neuro Oncol 2019 01;21(Suppl 1):i18-i31

Department of Neurological Surgery, University of California, San Francisco, California, USA.

Surgery has long been established as the first-line treatment for the majority of symptomatic and enlarging meningiomas, and evidence for its success is derived from retrospective case series. Despite surgical resection, a subset of meningiomas display aggressive behavior with early recurrences that are difficult to treat. The decision to radically resect meningiomas and involved structures is balanced against the risk for neurological injury in patients. Radiation therapy has largely been used as a complementary and safe therapeutic strategy in meningiomas with evidence primarily stemming from retrospective, single-institution reports. Two of the first cooperative group studies (RTOG 0539 and EORTC 22042) evaluating the outcomes of adjuvant radiation therapy in higher-risk meningiomas have shown promising preliminary results. Historically, systemic therapy has resulted in disappointing results in meningiomas. However, several clinical trials are under way evaluating the efficacy of chemotherapies, such as trabectedin, and novel molecular agents targeting Smoothened, AKT1, and focal adhesion kinase in patients with recurrent meningiomas.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1093/neuonc/noy136DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6347080PMC
January 2019

Ultraviolet light-related DNA damage mutation signature distinguishes cutaneous from mucosal or other origin for head and neck squamous cell carcinoma of unknown primary site.

Head Neck 2019 06 11;41(6):E82-E85. Epub 2019 Jan 11.

Department of Anatomic Pathology, University of California San Francisco, San Francisco, California.

Background: Head and neck squamous cell carcinoma of unknown primary site (HNSCCUP) is a diagnostic challenge. Identification of an ultraviolet (UV) light-related DNA damage signature using next-generation sequencing (NGS) can classify the primary site of origin as cutaneous.

Methods: A 62-year-old male was seen with 2 months of left neck swelling. He was a lifetime nonsmoker but had a history of cutaneous squamous cell carcinoma (SCC) of the left helix. He was also found to have left hilar adenopathy. He had a p16-negative HNSCCUP on fine needle aspiration (FNA) biopsy of the left neck.

Results: NGS of the FNA specimen revealed a high number of somatic mutations that were mostly C to T transitions, indicating a UV mutation signature and confirming the diagnosis of cutaneous SCC.

Conclusions: Identification of a UV DNA damage signature with NGS distinguishes HNSCCUP of cutaneous vs mucosal or other squamous cell carcinoma origin.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1002/hed.25613DOI Listing
June 2019

The application of artificial intelligence in the IMRT planning process for head and neck cancer.

Oral Oncol 2018 12 31;87:111-116. Epub 2018 Oct 31.

Department of Radiation Oncology, University of California, San Francisco, CA 94115, USA. Electronic address:

Artificial intelligence (AI) is beginning to transform IMRT treatment planning for head and neck patients. However, the complexity and novelty of AI algorithms make them susceptible to misuse by researchers and clinicians. Understanding nuances of new technologies could serve to mitigate potential clinical implementation pitfalls. This article is intended to facilitate integration of AI into the radiotherapy clinic by providing an overview of AI algorithms, including support vector machines (SVMs), random forests (RF), gradient boosting (GB), and several variations of deep learning. This document describes current AI algorithms that have been applied to head and neck IMRT planning and identifies rapidly growing branches of AI in industry that have potential applications to head and neck cancer patients receiving IMRT. AI algorithms have great clinical potential if used correctly but can also cause harm if misused, so it is important to raise the level of AI competence within radiation oncology so that the benefits can be realized in a controlled and safe manner.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.oraloncology.2018.10.026DOI Listing
December 2018

DoseNet: a volumetric dose prediction algorithm using 3D fully-convolutional neural networks.

Phys Med Biol 2018 12 4;63(23):235022. Epub 2018 Dec 4.

These two authors contributed equally.

The goal of this study is to demonstrate the feasibility of a novel fully-convolutional volumetric dose prediction neural network (DoseNet) and test its performance on a cohort of prostate stereotactic body radiotherapy (SBRT) patients. DoseNet is suggested as a superior alternative to U-Net and fully connected distance map-based neural networks for non-coplanar SBRT prostate dose prediction. DoseNet utilizes 3D convolutional downsampling with corresponding 3D deconvolutional upsampling to preserve memory while simultaneously increasing the receptive field of the network. DoseNet was implemented on 2 Nvidia 1080 Ti graphics processing units and utilizes a 3 phase learning protocol to help achieve convergence and improve generalization. DoseNet was trained, validated, and tested with 151 patients following Kaggle completion rules. The dosimetric quality of DoseNet was evaluated by comparing the predicted dose distribution with the clinically approved delivered dose distribution in terms of conformity index, heterogeneity index, and various clinically relevant dosimetric parameters. The results indicate that the DoseNet algorithm is a superior alternative to U-Net and fully connected methods for prostate SBRT patients. DoseNet required ~50.1 h to train, and ~0.83 s to make a prediction on a 128  ×  128  ×  64 voxel image. In conclusion, DoseNet is capable of making accurate volumetric dose predictions for non-coplanar SBRT prostate patients, while simultaneously preserving computational efficiency.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1088/1361-6560/aaef74DOI Listing
December 2018

Reducing radiation-related morbidity in the treatment of nasopharyngeal carcinoma.

Future Oncol 2017 Feb 23;13(5):425-431. Epub 2016 Nov 23.

Department of Radiation Oncology, University of California San Francisco, San Francisco, CA, USA.

While radiation therapy is the mainstay of treatment for nasopharyngeal carcinoma, the anatomic location of the nasopharynx in close proximity to radiation-sensitive organs such as the salivary glands, optic nerves and chiasm, cochlea, brainstem and temporal lobes presents a special challenge. Technological approaches to reducing the morbidity of nasopharyngeal cancer irradiation have been historically successful with the evolution from 2D techniques to increasingly conformal forms of radiation therapy. This report reviews normal tissue dose constraints and major considerations in target delineation for patients with nasopharyngeal cancer in the intensity-modulated radiation therapy era. Furthermore, this report discusses more contemporary approaches to toxicity reduction such as the judicious reduction or omission of radiation to low-risk regions and the potential role of particle beam therapy.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.2217/fon-2016-0410DOI Listing
February 2017

UVA phototransduction drives early melanin synthesis in human melanocytes.

Curr Biol 2011 Nov 3;21(22):1906-11. Epub 2011 Nov 3.

Department of Molecular Pharmacology, Physiology and Biotechnology, Brown University, Providence, RI 02912, USA.

Exposure of human skin to solar ultraviolet radiation (UVR), a powerful carcinogen [1] comprising ~95% ultraviolet A (UVA) and ~5% ultraviolet B (UVB) at the Earth's surface, promotes melanin synthesis in epidermal melanocytes [2, 3], which protects skin from DNA damage [4, 5]. UVB causes DNA lesions [6] that lead to transcriptional activation of melanin-producing enzymes, resulting in delayed skin pigmentation within days [7]. In contrast, UVA causes primarily oxidative damage [8] and leads to immediate pigment darkening (IPD) within minutes, via an unknown mechanism [9, 10]. No receptor protein directly mediating phototransduction in skin has been identified. Here we demonstrate that exposure of primary human epidermal melanocytes (HEMs) to UVA causes calcium mobilization and early melanin synthesis. Calcium responses were abolished by treatment with G protein or phospholipase C (PLC) inhibitors or by depletion of intracellular calcium stores. We show that the visual photopigment rhodopsin [11] is expressed in HEMs and contributes to UVR phototransduction. Upon UVR exposure, significant melanin production was measured within one hour; cellular melanin continued to increase in a retinal- and calcium-dependent manner up to 5-fold after 24 hr. Our findings identify a novel UVA-sensitive signaling pathway in melanocytes that leads to calcium mobilization and melanin synthesis and may underlie the mechanism of IPD in human skin.
View Article and Find Full Text PDF

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.cub.2011.09.047DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3586554PMC
November 2011