ABSTRACT
Machine learning has become a key driver of the digital health revolution. That comes with a fair share of high hopes and hype. We conducted a scoping review on machine learning in medical imaging, providing a comprehensive outlook of the field's potential, limitations, and future directions. Most reported strengths and promises included: improved (a) analytic power, (b) efficiency (c) decision making, and (d) equity. Most reported challenges included: (a) structural barriers and imaging heterogeneity, (b) scarcity of well-annotated, representative and interconnected imaging datasets (c) validity and performance limitations, including bias and equity issues, and (d) the still missing clinical integration. The boundaries between strengths and challenges, with cross-cutting ethical and regulatory implications, remain blurred. The literature emphasizes explainability and trustworthiness, with a largely missing discussion about the specific technical and regulatory challenges surrounding these concepts. Future trends are expected to shift towards multi-source models, combining imaging with an array of other data, in a more open access, and explainable manner.
ABSTRACT
BACKGROUND: Preclinical models show that an antiangiogenic regimen at low-dose daily (metronomic) dosing may be effective against chemotherapy-resistant tumors. We undertook a prospective, open-label, single-arm, multi-institutional phase II study to evaluate the efficacy of a "5-drug" oral regimen in children with recurrent or progressive cancer. PROCEDURE: Patients ≤21 years old with recurrent or progressive tumors were eligible. Treatment consisted of continuous oral celecoxib, thalidomide, and fenofibrate, with alternating 21-day cycles of low-dose cyclophosphamide and etoposide. Primary endpoint was to assess, within eight disease strata, activity of the 5-drug regimen over 27 weeks. Blood and urine angiogenesis markers were assessed. RESULTS: One hundred one patients were enrolled; 97 began treatment. Median age was 10 years (range: 191 days-21 years); 47 (49%) were female. Disease strata included high-grade glioma (HGG, 21 patients), ependymoma (19), low-grade glioma (LGG, 12), bone tumors (12), medulloblastoma/primitive neuroectodermal tumor (PNET, 8), leukemia (4), neuroblastoma (3), and miscellaneous tumors (18). Treatment was generally well tolerated; most common toxicities were hematologic. Twenty-four (25%) patients completed 27 weeks therapy without progression, including HGG: 1 (5%), ependymoma: 7 (37%), LGG: 7 (58%), medulloblastoma/PNET: 1, neuroblastoma: 1, and miscellaneous tumors: 7 (39%). Best response was complete response (one patient with medulloblastoma), partial response (12), stable disease (36), progressive disease (47), and inevaluable (1). Baseline serum thrombospondin levels were significantly higher in patients successfully completing therapy than in those who progressed (P = 0.009). CONCLUSION: The 5-drug regimen was well tolerated. Clinical activity was demonstrated in some but not all tumor strata.
