In patients with intracerebral hemorrhage (ICH), direct lacerations of brain tissue cause immediate and delayed neurological damage that may result in death and long-term disability. To date, therapeutic approaches focus primarily on preventing ICH expansion by controlling blood pressure, reversing any associated coagulopathy, and providing care in dedicated stroke unit [1]. During previous years, hemostatic agents in the absence of known coagulopathy have attracted interest but randomized clinical trials failed to demonstrate their benefit [1]. Most recently, the Early Minimally Invasive Removal of Intracerebral Hemorrhage (ENRICH) clinical trial has raised hopes by demonstrating, for the first time, a functional benefit of minimally invasive surgical drainage in supra-tentorial ICH [2]. Another strategy under development is to target secondary injury pathways. In the days and weeks following the initial injury, the cytotoxic release and neuroinflammatory response associated with ICH cause further damage to surrounding tissue, which may be inhibited by drugs [3]. The evaluation of these novel therapies will require biomarkers to select patients most at risk of secondary brain injury. Because the diagnosis of ICH relies primarily on computed tomography (CT) examination, CT-based biomarkers would be ideal for this purpose. As a site receiving red cell debris and inflammatory mediators, peri‑ICH edema seems to be the most promising location to find such biomarkers [3]. In this issue of Diagnostic and Interventional Imaging, Huang et al. performed a CT-based radiomics analysis thoughtfully focusing on peri‑ICH edema [4]. They built and validated a radiomics based-CT model predictive of functional outcome at 90-day using admission noncontrast CT examination. Their radiomics analysis (i.e., the quantification of images texture and shape characteristics) indicates a greater heterogeneity of peri‑ICH edema in patients with worst functional outcome, thereby suggesting a relationship with the occurrence of secondary brain injury in these patients. Translational studies on animal models could strengthen the interpretability of the radiomics parameters highlighted in this study by associating them specifically with the various histopathological lesions observed in the peri‑ICH zone [5]. However, as acknowledged by the authors, the clinical value added by this complex and time-consuming approach remains modest at present, when compared to a conventional model that considers easily accessible parameters such as patient age, severity of consciousness impairment according to the Glasgow Coma Scale, blood glucose, and ICH volume and location. Prospects for improvement undoubtedly include optimizing the timing of data acquisition. The development of peri‑ICH edema is a highly dynamic process that spans several weeks, reaching its peak volume around the third week [3]. Throughout this period, a cascade of events occurs, involving excitotoxicity, blood-brain barrier disruption, and neuroinflammation. While the primary goal of this cascade is ICH resorption, its dysregulation could have adverse consequences for some patients. In this context, assessing the pathological mechanisms at play in the peri‑ICH zone within the first six hours of ICH formation may be too early to capture the dynamics accurately. Ultimately, for the future use of such a tool in clinical practice, two key requirements need to be addressed. They include validation of its robustness against local technical variations, including machine effects and segmentation methods, and smooth integration into the radiologist's workflow, encompassing the full automation of the entire procedure (including peri‑ICH edema segmentation) [6]. Only under these conditions will radiomics offer a new perspective on peri‑ICH edema and guide the utilization of innovative therapies.