Clinical Evaluation of Deep Learning for Tumor Delineation on 18F-FDG PET/CT of Head and Neck Cancer.

Artificial intelligence (AI) may decrease 18F-FDG PET/CT-based gross tumor volume (GTV) delineation variability and automate tumor-volume-derived image biomarker extraction. Hence, we aimed to identify and evaluate promising state-of-the-art deep learning methods for head and neck cancer (HNC) PET GTV delineation. Methods: We trained and evaluated deep learning methods using retrospectively included scans of HNC patients referred for radiotherapy between January 2014 and December 2019 (ISRCTN16907234). We used 3 test datasets: an internal set to compare methods, another internal set to compare AI-to-expert variability and expert interobserver variability (IOV), and an external set to compare internal and external AI-to-expert variability. Expert PET GTVs were used as the reference standard. Our benchmark IOV was measured using the PET GTV of 6 experts. The primary outcome was the Dice similarity coefficient (DSC). ANOVA was used to compare methods, a paired t test was used to compare AI-to-expert variability and expert IOV, an unpaired t test was used to compare internal and external AI-to-expert variability, and post hoc Bland-Altman analysis was used to evaluate biomarker agreement. Results: In total, 1,220 18F-FDG PET/CT scans of 1,190 patients (mean age ± SD, 63 ± 10 y; 858 men) were included, and 5 deep learning methods were trained using 5-fold cross-validation (n = 805). The nnU-Net method achieved the highest similarity (DSC, 0.80 [95% CI, 0.77-0.86]; n = 196). We found no evidence of a difference between expert IOV and AI-to-expert variability (DSC, 0.78 for AI vs. 0.82 for experts; mean difference of 0.04 [95% CI, -0.01 to 0.09]; P = 0.12; n = 64). We found no evidence of a difference between the internal and external AI-to-expert variability (DSC, 0.80 internally vs. 0.81 externally; mean difference of 0.004 [95% CI, -0.05 to 0.04]; P = 0.87; n = 125). PET GTV-derived biomarkers of AI were in good agreement with experts. Conclusion: Deep learning can be used to automate 18F-FDG PET/CT tumor-volume-derived imaging biomarkers, and the deep-learning-based volumes have the potential to assist clinical tumor volume delineation in radiation oncology.

Repeatability of FDG PET/CT metrics assessed in free breathing and deep inspiration breath hold in lung cancer patients.

We measured the repeatability of FDG PET/CT uptake metrics when acquiring scans in free breathing (FB) conditions compared with deep inspiration breath hold (DIBH) for locally advanced lung cancer. Twenty patients were enrolled in this prospective study. Two FDG PET/CT scans per patient were conducted few days apart and in two breathing conditions (FB and DIBH). This resulted in four scans per patient. Up to four FDG PET avid lesions per patient were contoured. The following FDG metrics were measured in all lesions and in all four scans: Standardized uptake value (SUV)peak, SUVmax, SUVmean, metabolic tumor volume (MTV) and total lesion glycolysis (TLG), based on an isocontur of 50% of SUVmax. FDG PET avid volumes were delineated by a nuclear medicine physician. The gross tumor volumes (GTV) were contoured on the corresponding CT scans. Nineteen patients were available for analysis. Test-retest standard deviations of FDG uptake metrics in FB and DIBH were: SUVpeak FB/DIBH: 16.2%/16.5%; SUVmax: 18.2%/22.1%; SUVmean: 18.3%/22.1%; TLG: 32.4%/40.5%. DIBH compared to FB resulted in higher values with mean differences in SUVmax of 12.6%, SUVpeak 4.4% and SUVmean 11.9%. MTV, TLG and GTV were all significantly smaller on day 1 in DIBH compared to FB. However, the differences between metrics under FB and DIBH were in all cases smaller than 1 SD of the day to day repeatability. FDG acquisition in DIBH does not have a clinically relevant impact on the uptake metrics and does not improve the test-retest repeatability of FDG uptake metrics in lung cancer patients.

Spatio-temporal stability of pre-treatment 18F-Fludeoxyglucose uptake in head and neck squamous cell carcinomas sufficient for dose painting.

BACKGROUND: The pre-treatment 18F-Fludeoxyglucose (FDG) avid subvolume of the tumor has shown promise as a potential target for dose painting in patients with in head and neck squamous cell carcinomas (HNSCC). PURPOSE: The purposes of this study are: 1) to assess the pre-treatment spatio-temporal variability of FDG PET/CT target volumes and 2) to assess the impact of this variability on dose distribution in dose painting plans in patients with HNSCC. MATERIAL AND METHODS: Thirty patients were enrolled and scanned twice, three days apart, days prior to treatment. Delineation of the FDG avid subvolume of the tumor and lymph nodes on both scans was performed by a specialist in nuclear medicine yielding GTVPET1 and GTVPET2 and segmentation based on SUV iso-contours were constructed yielding two metabolic target volumes, MTV1 and MTV2. Images were co-registered rigidly and dose painting plans with dose escalation up to 82 Gy to GTVPET1 were planned and GTVPET2 was copied from the co-registered images to the dose planning scan. Variation in dose to the target and modeled tumor control probability were assessed as measures of the impact of imaging variations in a dose painting scenario. RESULTS: Twenty-four patients were available for full analysis. The median mismatch between GTVPET1 and GTVPET2 was 14.2% (1.7 cm(3)). The median difference in dose to the FDG planning target volume was 0.3 Gy (PTVPET) and 0.4 Gy (PTVMTV). Median difference in the modeled tumor control probability (TCP) was < 0.2% and 23 of 24 patients had a difference in expected TCP < 1%. CONCLUSIONS: Pre-treatment FDG PET/CT target volumes were stable and day-to-day variability had no relevant impact on dose distribution and expected tumor control in dose painting plans.

A prospective comparison of 18F-FDG PET/CT and CT as diagnostic tools to identify the primary tumor site in patients with extracervical carcinoma of unknown primary site.

BACKGROUND: The aim of the present study was to evaluate prospectively the diagnostic value of (18)F-fluorodeoxyglucose positron emission tomography/computed tomography ((18)F-FDG PET/CT) and conventional CT regarding the ability to detect the primary tumor site in patients with extracervical metastases from carcinoma of unknown primary (CUP) site. PATIENTS AND METHODS: From January 2006 to December 2010, 136 newly diagnosed CUP patients with extracervical metastases underwent (18)F-FDG PET/CT. A standard of reference (SR) was established by a multidisciplinary team to ensure that the same set of criteria were used for classification of patients, that is, either as CUP patients or patients with a suggested primary tumor site. The independently obtained suggestions of primary tumor sites using PET/CT and CT were correlated with the SR to reach a consensus regarding true-positive (TP), true-negative, false-negative, and false-positive results. RESULTS: SR identified a primary tumor site in 66 CUP patients (48.9%). PET/CT identified 38 TP primary tumor sites and CT identified 43 TP primary tumor sites. No statistically significant differences were observed between (18)F-FDG PET/CT and CT alone in regard to sensitivity, specificity, and accuracy. CONCLUSION: In the general CUP population with multiple extracervical metastases (18)F-FDG PET/CT does not represent a clear diagnostic advantage over CT alone regarding the ability to detect the primary tumor site.

Immune reconstitution syndrome presenting as probable AIDS-related lymphoma: a case report.

We report an unusual case of HIV-related immune reconstitution inflammatory syndrome, presenting as suspected AIDS-related lymphoma. Symptoms, initial investigations including fine-needle biopsy and 18F-FDG PET/CT scan were highly compatible with high grade AIDS-related lymphoma, however subsequently IRIS was diagnosed. We discuss pitfalls in the interpretation of diagnostic results in ARL versus IRIS.