Bone metastases represent a frequent and serious complication in patients with advanced solid tumours, commonly arising in cancers of the prostate, breast and lung [1]. These metastatic lesions can trigger skeletal-adverse events (SREs), such as debilitating pain, restricted mobility, spinal cord and nerve root compression, myelosuppression, pathologic fractures, and hypercalcemia [2]. Beyond their direct impact on prognosis and patient morbidity [3], SREs significantly reduce patients’ health-related quality of life and autonomy [4]. Early detection of metastases is thus paramount to enable timely interventions (i.e., radiotherapy, surgical fixation, chemotherapy) and the prevention of serious outcomes [5]. 

Current diagnostic protocols rely heavily on imaging, with computed tomography (CT) playing a pivotal role for the initial staging and follow-up of cancer. It offers fine anatomical resolution and soft-tissue contrast that allow for the detection of small and subtle bone lesions [6, 7, 8]. Despite the accuracy and cost-effectiveness of CT scanning, identifying bone metastases remains a labor-intensive task that is prone to error [9]. First, radiologists must scrutinize multiple anatomical regions across hundreds of images, often under various window settings. Second, sclerotic metastases can be difficult to differentiate from common benign abnormalities such as bone islands. Third, many scans are not primarily indicated for cancer follow-up, increasing the risk of oversights due to the important time constraints faced by radiologists. 

Automating this task with artificial intelligence (AI) holds significant promise for improving the detection and management of bone metastases, as highlighted by the Data Science Institute of the American College of Radiology. Despite proven efficacy of AI in the diagnosis and staging of cancers like those of the lung and breast, focused research on bone metastases remains scarce [10, 11]. This study aims to address this gap by evaluating the performance of a novel AI tool for the detection of osteolytic and sclerotic bone metastases on CT scans, and comparing it against that of experienced radiologists.