Federated learning enables big data for rare cancer boundary detection

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.

External Beam Irradiation and Restenosis Following Femoral Stenting: Long-Term Results of a Prospective Randomized Study

PURPOSE

To assess the long-term outcome of external beam irradiation (EBI) for the prevention of restenosis due to neointimal hyperplasia, following percutaneous transluminal angioplasty (PTA) and stenting of the superficial femoral artery.

METHODS

Sixty consecutive patients with peripheral arterial disease, who were treated with "bail-out" stent implantation in the superficial femoral artery due to suboptimal PTA, were included in this study. Patients were randomly allocated into two groups, receiving either external beam irradiation (6 MV photons, total dose 24 Gy in a hypofractionated schedule) plus antiplatelet therapy (EBI group) or antiplatelet therapy alone (control group).

RESULTS

No procedure-related complications occurred, and all patients of the EBI group received the full dose of 24 Gy. During the long-term follow-up, an overall statistically significant difference was demonstrated in favor of the EBI group patients, regarding both the in-stent (log-rank test, p = 0.0072) and the in-segment binary restenosis (log-rank test, p = 0.0103). The primary patency rates were also significantly better in the EBI group at specific time-points, such as in the first (74.2% vs 46.5%, p = 0.019), second (62.5% vs 33.8%, p = 0.020), and third (54.6% vs 29.0%, p = 0.039) year, respectively. Moreover, the overall clinically driven reintervention rate was significantly lower among patients of the irradiated group (log-rank test, p = 0.038).

CONCLUSION

Our long-term follow-up analysis revealed that EBI following femoral artery PTA and stenting significantly reduces restenosis and reintervention rates, while improving primary patency.

External beam radiation therapy reduces the rate of re-stenosis in patients treated with femoral stenting: results of a randomised study

BACKGROUND AND PURPOSE

To evaluate the feasibility and efficacy of external beam irradiation (EBI) for the prevention of re-stenosis due to neointimal hyperplasia, after percutaneous transluminal angioplasty (PTA) and stent placement of the superficial femoral artery.

PATIENTS AND METHODS

A total of 60 patients with the diagnosis of superficial femoral artery stenoses or occlusions due to peripheral arterial obstructive disease underwent PTA and implantation of a self-expandable stent at their superficial femoral artery. After the procedure, patients were randomised and 30 of them received EBI (6 MV photons, total dose 24 Gy in six fractions in 2 weeks), while the rest 30 received no radiation therapy.

RESULTS

EBI was technically feasible in all patients, without serious radiation related side effects. Overall, a statistically significant difference was observed in stenosis categories between the two groups at 6 months follow-up (P=0.04). More specifically, significantly more patients in the control group presented with stenosis greater or equal than 70% [EBI group 30% (9/30); control group 66.7% (20/30); P=0.009]. This difference in the percentage of re-stenosis had as a consequence significantly lower re-intervention rates among the patients of the irradiated group [17% (5/30) versus 47% (14/30); P=0.025] during the 6 months follow-up period. We also observed that the irradiated patients had re-stenosis at the stent ends, while the non-irradiated had re-stenosis at the stent ends and the lumen. Three of the irradiated patients, who discontinued the anti-platelet treatment, have shown thrombosis of the irradiated artery during the first month from the completion of the treatment.

CONCLUSIONS

It is our belief that EBI is a feasible, safe and effective method for the prevention of neointimal hyperplasia at the superficial femoral artery. Further studies are deemed necessary to optimise the radiotherapy schedule.