الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 106 |  |  |
| | | from | 106 | | |
Abstract
Head and neck cancer is the eighth leading cause of cancer-related death worldwide. Head and neck squamous cell carcinoma (HNSCC) accounts for more than 90% of head and neck cancers. For early stage (stage I-II) HNSCC, a single modality treatment with surgery or radiation therapy is generally recommended; while combined modality therapy is generally needed for patients with advanced disease (stage III–IV) at diagnosis. Therefore, accurate preoperative staging of HNSCC is essential for developing the most proper treatment strategy, which has an important impact on both prognosis and quality of life(1). As regard Thyroid cancer, it represents the most common malignancy of the endocrine system with an increasing incidence worldwide. Most thyroid malignancies are well-differentiated carcinomas, namely, papillary and follicular carcinoma. Medullary and undifferentiated carcinomas are rarer and account for 5% and 2% of all thyroid cancers, respectively(2). Machine learning (ML) includes a broad class of computer programs that grow with experience. The difficulty of generating, training, and checking machine learning indicates that the success of the algorithms will require radiologist involvement for years to come, leading to engagement rather than replacement. It is difficult to ignore the increasing interest in machine learning. ML algorithms generate public interest because they include playing games against humans, self-driving cars, and identifying the characteristics of a great selfie(3). In radiology, Recent progresses and future perspectives of machine learning techniques offer hopeful applications in medical imaging, It has the potential to improve different steps of the radiology workflow including order scheduling and triage, clinical decision support systems, detection and interpretation of findings, postprocessing and dose estimation, examination quality control, and radiology reporting(4). In conclusion, our study indicates, using the current state of the art machine learning approaches, the potential of multi-parametric (DWI and structural T2) MR images for noninvasively categorizing the malignancy of thyroid nodules in a single, definitive procedure, thus sparing patients from unnecessary operations and waiting times associated with a diagnostic lobectomy. There was no significant diagnostic difference between different b-values. However, the fusion of functional, morphology as well as appearance imaging and volumetric markers has demonstrated higher accuracy compared with individual image markers. In the future we plan to expand our analysis using larger datasets as well as including other clinical biomarkers.