151751-Najiba-Chargi

432 CHAPTER 21 Cespedes Feliciano et al. recently published a promising study in which skeletal muscle mass and adipose tissue was automatically segmented in patients with non-metastatic colorectal (n=3102) and breast cancer (n=2888) at the level of L3 using automated software. 36 They also performedmanual skeletal musclemass and adipose-tissue segmentations. There was strong agreement between manual and automatic segmentations overall and within subgroups of age, sex, body mass index, and cancer stage: average Jaccard scores and intra-class correla - tion coefficients exceeded 90% for all tissues. However, the authors describe that automated segmentation performance was lowest for the <2% of patients who were underweight or had anatomic abnormalities. Head and neck cancer patients are more likely to have low body weight at diagnosis than patients with colorectal cancer or breast cancer, therefore further studies investigating the implication of automatic segmentation of body composition in head and neck cancer patients are warranted. Another study performed by Edwards et al. used deep learning, fully convolutional neural network for the segmentation of abdominal muscle on CT and showed a mean Dice similarity coefficient of 0.92, a mean precision of 0.93, and a mean recall of 0.91 in an independent test set. 37 Another study performed by Blanc-Durand et al. in 189 patients with lung cancer showed that deep-learning was able to distinguish subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and muscular body mass (MBM) with mean Dice similarity coefficients in the validation set were 0.95, 0.93, and 0.91 for SAT, VAT, and MBM, respectively. 38  The by deep-learning obtained BSA-normalized VAT/SAT ratio was shown to be an independent predictor for survival in lung cancer patients. Currently, no AI studies for body composition are performed in head and neck cancer patients. CONCLUSION In conclusion, in head and neck cancer patients, low skeletal muscle mass is a prevalent problem which occurs in approximately 55% of patients. Skeletal muscle mass can be easily assessed on a single slice at the level of C3 (or L3) on routinely performed CT or MRI scans which are performed for head and neck cancer diagnosis and treatment evaluation. Skeletal muscle mass is a promising as imaging biomarker which predicts negative treatment out- comes in various treatment strategies applied in head and neck cancer management. Besides negative treatment outcomes, low skeletal muscle mass has also shown to be prognostic for decreased survival. We hypothesize that multimodal pre-habilitation will improve skeletal muscle mass status of the patient before treatment which will lead to an enhanced recovery trajectory with reduced operative complications and postoperative adverse effects in surgically treated patients and to reduced treatment-related toxicities in patients treated with (chemo- or bio) radiotherapy. We also hypothesize that multimodal pre-habilitation leads to a reduced duration of hospital stay, reduced health care costs and improved quality of life. In addition, pre-habilitation is an opportunity to foster patient empowerment which increases patient’s autonomy and self-man - agement. This may facilitate an improved quality of life before treatment and may positively affect long-term health. Therefore, the aims of a future randomized controlled trial should be to compare the effect of a multimodal pre-habilitation program including exercise, nutritional