Zeno A.R. Gouw

Zeno A.R. Gouw

Auto-generated profile, based on publicly available information

Zeno A.R. Gouw

Memorial Sloan Kettering Cancer Center, Department of Medical Physics, New York, NY, USA; The Netherlands Cancer Institute, Amsterdam, Department of Radiation Oncology, the Netherlands

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 June 2021 in European Journal of Radiology
Verified
Journal Article
Closed access
PBPaula Bos
Paula Bos
Michiel Van den BrekelMichiel Van den Brekel
Michiel Van den Brekel
ZGZeno A.R. Gouw
Zeno A.R. Gouw
AAAbrahim Al-Mamgani
Abrahim Al-Mamgani
MTMarjaneh Taghavi
Marjaneh Taghavi
BJBas Jasperse
Bas Jasperse

5 August 2024 in Physics in Medicine & Biology
Verified
Journal Article
Open access
JRJintao Ren
Jintao Ren
JTJonas Teuwen
Jonas Teuwen
Jasper NijkampJasper Nijkamp
Jasper Nijkamp
MRMathis Rasmussen
Mathis Rasmussen
ZGZeno Gouw
Zeno Gouw
SKStine Korreman
Stine Korreman
Objective. Deep learning shows promise in autosegmentation of head and neck cancer (HNC) primary tumours (GTV-T) and nodal metastases (GTV-N). However, errors such as including non-tumour regions or missing nodal metastases still occur. Conventional methods often make overconfident predictions, compromising reliability. Incorporating uncertainty estimation, which provides calibrated confidence intervals can address this issue. Our aim was to investigate the efficacy of various uncertainty estimation methods in improving segmentation reliability. We evaluated their confidence levels in voxel predictions and ability to reveal potential segmentation errors. Approach. We retrospectively collected data from 567 HNC patients with diverse cancer sites and multi-modality images (CT, PET, T1-, and T2-weighted MRI) along with their clinical GTV-T/N delineations. Using the nnUNet 3D segmentation pipeline, we compared seven uncertainty estimation methods, evaluating them based on segmentation accuracy (Dice similarity coefficient, DSC), confidence calibration (Expected Calibration Error, ECE), and their ability to reveal segmentation errors (Uncertainty-Error overlap using DSC, UE-DSC). Main results. Evaluated on the hold-out test dataset (n = 97), the median DSC scores for GTV-T and GTV-N segmentation across all uncertainty estimation methods had a narrow range, from 0.73 to 0.76 and 0.78 to 0.80, respectively. In contrast, the median ECE exhibited a wider range, from 0.30 to 0.12 for GTV-T and 0.25 to 0.09 for GTV-N. Similarly, the median UE-DSC also ranged broadly, from 0.21 to 0.38 for GTV-T and 0.22 to 0.36 for GTV-N. A probabilistic network—PhiSeg method consistently demonstrated the best performance in terms of ECE and UE-DSC. Significance. Our study highlights the importance of uncertainty estimation in enhancing the reliability of deep learning for autosegmentation of HNC GTV. The results show that while segmentation accuracy can be similar across methods, their reliability, measured by calibration error and uncertainty-error overlap, varies significantly. Used with visualisation maps, these methods may effectively pinpoint uncertainties and potential errors at the voxel level.

1 November 2020 in International Journal of Radiation Oncology - Biology - Physics (IJROBP)
Verified
Journal Article
Closed access
ZGZeno A.R. Gouw
Zeno A.R. Gouw
MFMatthew D. La Fontaine
Matthew D. La Fontaine
WVWouter V. Vogel
Wouter V. Vogel
JKJeroen B. van de Kamer
Jeroen B. van de Kamer
JSJan-Jakob Sonke
Jan-Jakob Sonke
AAAbrahim Al-Mamgani
Abrahim Al-Mamgani

10 July 2023 in Radiotherapy and Oncology
Verified
Journal Article
Open access
ABAnne L.H. Bisgaard
Anne L.H. Bisgaard
RKRick Keesman
Rick Keesman
ALAstrid L.H.M.W. van Lier
Astrid L.H.M.W. van Lier
CCCatherine Coolens
Catherine Coolens
PHPetra J. van Houdt
Petra J. van Houdt
FMFaisal Mahmood
Faisal Mahmood
The apparent diffusion coefficient (ADC), a potential imaging biomarker for radiotherapy response, needs to be reproducible before translation into clinical use. The aim of this study was to evaluate the multi-centre delineation- and calculation-related ADC variation and give recommendations to minimize it. Nine centres received identical diffusion-weighted and anatomical magnetic resonance images of different cancerous tumours (adrenal gland, pelvic oligo metastasis, pancreas, and prostate). All centres delineated the gross tumour volume (GTV), clinical target volume (CTV), and viable tumour volume (VTV), and calculated ADCs using both their local calculation methods and each of the following calculation conditions: b-values 0-500 vs. 150-500 s/mm2, region-of-interest (ROI)-based vs. voxel-based calculation, and mean vs. median. ADC variation was assessed using the mean coefficient of variation across delineations (CVD) and calculation methods (CVC). Absolute ADC differences between calculation conditions were evaluated using Friedman's test. Recommendations for ADC calculation were formulated based on observations and discussions within the Elekta MRI-linac consortium image analysis working group. The median (range) CVD and CVC were 0.06 (0.02-0.32) and 0.17 (0.08-0.26), respectively. The ADC estimates differed 18% between b-value sets and 4% between ROI/voxel-based calculation (p-values < 0.01). No significant difference was observed between mean and median (p = 0.64). Aligning calculation conditions between centres reduced CVC to 0.04 (0.01-0.16). CVD was comparable between ROI types. Overall, calculation methods had a larger impact on ADC reproducibility compared to delineation. Based on the results, significant sources of variation were identified, which should be considered when initiating new studies, in particular multi-centre investigations.

1 October 2017 in Clinical and Translational Radiation Oncology
Verified
Journal Article
Open access
ZGZeno A.R. Gouw
Zeno A.R. Gouw
BJBas Jasperse
Bas Jasperse
JSJan-Jakob Sonke
Jan-Jakob Sonke
WHWilma D. Heemsbergen
Wilma D. Heemsbergen
AArash Navran
Arash Navran
AAAbrahim Al-Mamgani
Abrahim Al-Mamgani
Early detection of Residual disease (RD) is vital for salvage possibilities after (chemo) radiatiotherapy for oropharyngeal carcinoma (OPC). We standardized clinical investigation to test its added value to MRI response evaluation and investigated the benefit of FDG-PET/CT. Radiological response evaluation using Ojiri-score was done for 234 patients with OPC, using MRI 12 weeks after (chemo) radiotherapy between 2010 and 2014. The presence of mucosal lesions and/or major complaints (still completely tube feeding-dependent and/or opiate-dependent because of swallowing problems) was scored as clinical suspicion (CS). Retrospectively, the performance of Ojiri to predict RD was compared to CS and both combined using Pearson Chi-squared. Of the whole group, FDG-PET/CT metabolic response (MR) was available in 50 patients. Twelve out of 234 patients (5.1%) had RD. Ojiri and CS had excellent negative predictive value (NPV) (98% and 100% respectively). The combination of CS and Ojiri reduced false positives by 32% (38-26 patients) without lowering NPV (98%). No patients with complete MR (n = 39) at the FDG-PET/CT had RD compared to 5 (45%) with partial MR. For response evaluation in OPC, the combination of CS and Ojiri-score improved the predictive accuracy by reducing false positives compared to them individually. FDG-PET/CT is promising to further reduce false positives.

17 September 2023 in Clinical and Translational Radiation Oncology
Verified
Journal Article
Open access
AAAbrahim Al-Mamgani
Abrahim Al-Mamgani
RKRob Kessels
Rob Kessels
ZGZeno A.R. Gouw
Zeno A.R. Gouw
AArash Navran
Arash Navran
VMVineet Mohan
Vineet Mohan
WVWouter V. Vogel
Wouter V. Vogel
To report on the late toxicity and local control (LC) of head and neck cancer patients treated with adaptive FDG-PET/CT response-guided radiotherapy (ADMIRE) with dose escalation (NCT03376386). Between December 2017 and April 2019, 20 patients with stage II-IV squamous cell carcinoma of the larynx, hypopharynx or oropharynx were treated within the ADMIRE study where FDG-PET/CT response-guided (Week 2&4) dose escalation was applied (total dose 70–78 Gy). Cisplatin or cetuximab was added to radiotherapy in case of T3-4 and/or N2c disease. To compare the LC and late toxicity of the study population, we used an external control group (n = 67) consisting of all eligible patients for the study (but not participated). These patients were treated in our institution during the same period with the current standard of 70 Gy radiotherapy. To reduce the effect of confounding, logistic regression analyses was done using stabilized inverse probability of treatment weighting (SIPTW). After median follow-up of 40 and 43 months for the ADMIRE and control groups, the 3-year LC-rates were 74% and 78%, respectively (adjusted HR after SIPTW 0.80, 95 %CI 0.25–2.52, p = 0.70). The incidences of any late G3 toxicity were 35% and 18%, respectively. The adjusted OR for any late G3 toxicity was 5.09 (95 %CI 1.64–15.8, p = 0.005), for any late G ≥ 2 toxicity was 3.67 (95 %CI 1.2–11.7, p = 0.02), for persistent laryngeal edema was 10.95 (95% CI 2.71–44.29, p = 0.001), for persistent mucosal ulcers was 4.67 (95% CI 1.23–17.7, p = 0.023), and for late G3 radionecrosis was 15.69 (95 %CI 2.43–101.39, p = 0.004). Given the comparable LC rates with increased late toxicity in the ADMIRE group, selection criteria for future adaptive dose escalation trials (preferably randomized) need to be refined to include only patients at higher risk of local failure and/or lower risk of severe late toxicity.