Head and neck cancer is one of the most common cancers worldwide (fifth leading incidence). Radiotherapy and cetuximab are considered the standard treatment.
However, locoregional defects are a significant challenge in improving and controlling the condition and occur in forty percent of the patients within the first two years of the treatment.
Recently, several radiomic studies have been proposed to better identify patients in a non-invasive manner by using available images such as those obtained for diagnosis and planning treatment. These studies are based on positron emission tomography (PET) and computed tomography (CT) imaging.
HECKTOR was created in 2020 to provide an opportunity to compare automated algorithms dedicated to tumor segmentation in FDG PET/CT multidimensional images.
The HECKTOR Challenge is an excellent opportunity for professionals working on three-dimensional segmentation algorithms to present GTV tumor segmentation approaches in PET/CT scans, focusing on oral and pharyngeal cancers.
The 1st edition of the HECKTOR challenge was performed using a data set of 254 patients from 5 centers and attracted about 20 people to participate. In 2021, Hector’s challenge was repeated with a larger data set (325 patients from 6 centers) with the addition of a second task. The 2nd edition of the HECKTOR challenge was successfully held with twice the number of participants, 40 for the first task and 30 participants for the second task.
Following the success of the first two HECKTOR Challenges in 2020 and 2021, this challenge will again be held at the 25th International Conference on Medical Image Computing and Intervention (MICCAI) 2022.
This challenge version also includes fully automatic lymph node diagnosis and segmentation in addition to the primary tumor. The data set is expected to have more than a thousand patients collected from 9 different centers.
Hector Challenge is now hosted on the grand-challenge.org platform: hecktor.grand-challenge.org.
Why the HECKTOR Challenge?
Focusing on the metabolic and morphological characteristics of the tissue, the PET and CT methods include complementary and synergistic information for the classification of cancerous lesions, as well as tumor characteristics potentially relevant for predicting patient outcomes.
Modern image analysis methods must be developed and, more importantly, carefully evaluated to extract and apply this information.
The HECKTOR Challenge is organized to provide an opportunity for participants working on three-dimensional segmentation algorithms to develop their approaches to segmentation of primary tumors (GTVt) of the head and neck on PET/CT scans.
The scope of the challenge is expected to grow each year with a focus on a larger population (adding patients from new clinical centers) as well as adding new tasks or subdividing present tasks into sub-tasks.
Participation in HECKTOR
Participating teams will be allowed to have multiple submissions to evaluate their algorithms. Only the best submission from the teams will be considered as the results of the challenge. Each paper should be specifically identified and explained in the team paper sent along with the submissions.
Participants will not receive feedback before the deadline (except for errors).
Participants can use the data in any way they see fit. The use of additional data (public or non-public) is permitted but must be reported. Entrants who use the additional data will not be eligible for the HECKTOR challenge prize.
The results and winners will be announced publicly. Once participants submit their results to the challenge organizers via the challenge website, they will be fully considered for the challenge. Their results can be used in subsequent presentations, publications, or analyses at the company’s discretion.
Participating teams can publish their results and achievements on the LNCS. Participants can use the research results by citing the source, and no embargo will be imposed if they do so.
Submitting an article (minimum six pages, maximum 12 pages) is required to enter the official challenge and ranking.
If you or your team are interested in participating in more than one task, you are free to submit a paper that reports all methods and results or several papers. To link your paper to the team with which you participate in the Hecktor Challenge, you must add your team’s name at the end of the abstract. Successful teams will be contacted to deliver an oral presentation at MICCAI 2022.
TECVICO at HECKTOR
TECVICO actively takes part in HECKTOR as a team. TECVICO has a history of participating in the previous two editions of this challenge, winning prestigious awards.
Third Place for TECVICO at HECKTOR
TECVICO competed for the first time in the HECKTOR challenge as the Qurit Tecvico team, which resulted in being awarded 3rd place in the competition.
TECVICO participated in HECKTOR Challenge with “Advanced Automatic Segmentation of Tumors and Survival Prediction in Head and Neck Cancer” in 2021 and won 3rd place.
Positron emission tomography (PET) images were registered to computed tomography (CT) images, enhanced, and cropped. First, ten fusion techniques were utilized to combine PET and CT information. 3D-UNETR (UNET with Transformers) and 3D-UNET were utilized to automatically segment head and neck squamous cell carcinoma (HNSCC) tumors. Then 215 radiomic features from each region of interest via Standardized Environment for Radiomics Analysis (SERA) were extracted as part of this study.
Sixth Place for TECVICO at HECKTOR
TECVICO competed for the second time in the HECKTOR challenge as the Tecvico_Corp_Family team, which resulted in being awarded sixth place in the competition.
TECVICO participated in HECKTOR Challenge with “Fusion-Based Head and Neck Tumor Segmentation and Survival Prediction Using Robust Deep Learning Techniques and Advanced Hybrid Machine Learning Systems” in 2021 and won 6th place.
The article focuses on multi-level multi-modality fusion radiomics. It aims to employ advanced fusion techniques, deep learning segmentation methods, and survival analysis to automatically segment tumors and predict survival outcomes in head-and-neck-squamous-cell-carcinoma (HNSCC) cancer.
