Head and neck cancer (HNC) represents 3% of all malignancies, resulting in around 66,000 newly diagnosed cases and 15,000 deaths per year [1]. This disease is characterized by its heterogeneity at both histological and genetic levels and can originate from different anatomical regions such as the oral cavity, pharynx, and larynx HNC patients frequently exhibit cervical lymph node metastasis, and advanced-stage patients are prone to experience local recurrence [2] . Hypoxia is a crucial pathophysiological process in cancer biology, affecting the efficacy of radiation and drug therapy. While PET tracers like 18F-MISO and 64Cu-ATSM can provide varying degrees of hypoxia specificity, the downsides of PET, including ionizing radiation, poor spatial resolution, and the necessity for on-site cyclotron for radiotracer production, render these investigational PET imaging methods unfeasible in routine clinical settings [3]. Tissue oxygen concentration level dependent (TOLD) MRI with oxygen challenge is one of the potential MR imaging methods. This technique utilizes T1-weighted imaging to measure changes in tissue oxygen concentration levels induced by breathing room air compared to pure oxygen. Oxygen dissolved in blood and tissue plasma reduces the longitudinal relaxation time (T1), resulting in decreased T1 values in well‑oxygenated areas due to the paramagnetic effect of oxygen, while the T1 value of non‑oxygenated areas remains unchanged [4]. Several methods have been developed for quantitative T1 mapping, including techniques based on inversion recovery measurements, Look-Locker techniques [5,6], and variable flip angle methods [7,8]. Among these, spoiled gradient recalled echo (SPGRE) imaging with variable flip angles (VFA) is a popular technique for quickly mapping T1 relaxation time in vivo. This method offers superior noise efficiency compared to other T1 measurement methods and, when combined with quantitative flip angle mapping, can achieve competitive accuracy [8]. Most 3D VFA techniques utilize Cartesian sampling acquisitions, which make them prone to patient motion, particularly in the head and neck region. This susceptibility arises from the conventional Cartesian sampling method, which collects data along parallel lines in k-space. Patient motion during the scan can induce phase offsets, leading to image artifacts in the phase encoding direction and compromising T1 measurement accuracy [9]. To address this limitation, alternative non-cartesian trajectories, such as radial trajectories, have been proposed [[10], [11], [12]]. Achieving optimal accuracy in T1 measurements requires careful selection of pulse sequence parameters and acquisition sampling strategies. The hybrid 3D Radial and Cartesian sequence, also known as the 3D Stack of Stars (SOS) sequence has been used to acquire volumetric k-space data, where radial sampling is performed in-plane (along Ky and Kx) and Cartesian sampling is used along the slice dimension [13]. This k-space sampling scheme substantially reduces motion artifacts due to its varying sampling directions and oversampling of the center of the k-space [14]. To address motion sensitivity in 3D T1 mapping of head and neck cancer, we propose a technique employing a 3D stack-of-stars (SOS) acquisition. Although similar methods have been previously introduced in fast acquisition for temperature measurement [15], our approach relies on its inherent motion robustness, achieved through varied sampling directions and center k-space oversampling (14). To assess its impact on improving T1 measurements, we compare 3D T1 mapping utilizing the 3D SOS acquisition (15) with conventional Cartesian VFA acquisition, conducting studies in both phantom and head and neck cancer patients.