Deep learning to achieve clinically applicable segmentation of head and neck anatomy for radiotherapy

By Stanislav Nikolov, Sam Blackwell, R. Mendes, Jeffrey De Fauw, Clemens Meyer, Cían O. Hughes, Harry Askham, Bernardino Romera-Paredes, Alan Karthikesalingam, Carlton Chu and 13 more. arXiv.org, 2018

337 citations recorded by the source corpus.

Over half a million individuals are diagnosed with head and neck cancer each year worldwide. Radiotherapy is an important curative treatment for this disease, but it requires manually intensive delineation of radiosensitive organs at risk (OARs). This planning process can delay treatment commencement. While auto-segmentation algorithms offer a potentially time-saving solution, the challenges in defining, quantifying and achieving expert performance remain. Adopting a deep learning approach, we demonstrate a 3D U-Net architecture that achieves performance similar to experts in delineating a wide range of head and neck OARs. The model was trained on a dataset of 663 deidentified computed tomography (CT) scans acquired in routine clinical practice and segmented according to consensus OAR definitions. We demonstrate its generalisability through application to an independent test set of 24 CT scans available from The Cancer Imaging Archive collected at multiple international sites previously unseen to the model, each segmented by two independent experts and consisting of 21 OARs commonly segmented in clinical practice. With appropriate validation studies and regulatory approvals, this system could improve the effectiveness of radiotherapy pathways.

Subfields: Otorhinolaryngology, Radiation, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/cd557edc7bd4949c9ffb21a3e367ed90509a08fa

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Radiation

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

AbdomenCT-1K: Is Abdominal Organ Segmentation a Solved Problem? (2020, background): “Normalized surface Dice (NSD) [25], a boundary-based”
The Medical Segmentation Decathlon (2021, methodology): “The metrics DSCandNSD were chosen due to their popularity, rank stability [10], and smooth, well-understood and well-deﬁned behavior when ROIs do not overlap.”
Artificial intelligence and machine learning for medical imaging: A technology review. (2021, positive): “In numerous applications, CNNs have either demonstrated an accuracy similar to human experts [26,80,178,179], decreased the interobserver variability [180,181] or reduced the physician ’ s workload for a specific task [157] Fig.”
Self-derived organ attention for unpaired CT-MRI deep domain adaptation based MRI segmentation (2020, methodology): “…CT segmentation models can be trained using commonly available ‘off-the-shelf’ networks due to the availability of expert-labeled clinical datasets (Nikolov et al 2018, H ¨ ansch et al 2018, Zhu et al 2019), these methods cannot be directly applied to modalities like MRI because they lack…”
Modified U-Net (mU-Net) with Incorporation of Object-Dependent High Level Features for Improved Liver and Liver-Tumor Segmentation in CT Images (2019, methodology): “emerged and have been applied for liver [11]–[14], head and neck [15], [16], prostate [17], and brain [18], [19]”
CLIP-Driven Universal Model for Organ Segmentation and Tumor Detection (2023, background): “[49] Stanislav Nikolov, Sam Blackwell, Alexei Zverovitch, Ruheena Mendes, Michelle Livne, Jeffrey De Fauw, Yojan Patel, Clemens Meyer, Harry Askham, Bernardino Romera-Paredes, et al.”
High-performance medicine: the convergence of human and artificial intelligence (2019, background): “In contrast, the deep learning of the ECG pattern on the smartwatch, which can accurately detect whether there is high potassium in the blood, may provide particular usefulness for patients with kidney disease.”
TorchIO: A Python library for efficient loading, preprocessing, augmentation and patch-based sampling of medical images in deep learning (2020, background): “, 2017] or cuboids adapted to the anisotropic spacing of the training images [Nikolov et al., 2018].”
nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation (2020, methodology): “At the same semantic segmentation is an essential ingredient in numerous clinical applications 3,4 , including applications of artificial intelligence in diagnostic support systems 5,6 , therapy planning support 7 , intra-operative assistance 2 and tumor growth monitoring 8 .”
Self-Supervised Pre-Training of Swin Transformers for 3D Medical Image Analysis (2021, methodology): “The Dice similarity coefficient (Dice) and Hausdorff Distance 95% (HD95) are used as measurements for experiment results.”
AnatomyNet: Deep learning for fast and fully automated whole‐volume segmentation of head and neck anatomy (2018, neutral): “Its performance on this metric can be improved by considering surface and shape priors into the model as discussed above.(44,45)”
AMOS: A Large-Scale Abdominal Multi-Organ Benchmark for Versatile Medical Image Segmentation (2022, background): “Finally, we have also observed a signiﬁcant gap between mDice and mNSD as previously mentioned by [15].”
Medical Image Segmentation Review: The Success of U-Net (2022, background): “Medical image segmentation is a vital component and a cornerstone in most clinical applications, including diagnostic support systems [100], [101], therapy planning support [102], [103], intra-operative assistance [104], tumor growth monitoring [105], [106], and image-guided clinical surgery [107].”