Research Project
9622035
1 Abstract 2 3 This is a Phase I SBIR proposal to develop a non-contact and non-invasive imaging device for assisting clinicians 4 in selecting the appropriate level-of-amputation (LOA) in limbs with peripheral arterial disease (PAD). Surgeons 5 prefer to salvage as much limb tissue as possible during amputation to increase patient mobility while decreasing 6 morbidity and mortality. However, clinicians must balance this preference against the likelihood of primary wound 7 healing at a given level of amputation (LOA), which decreases with more distal amputations. There are no gold- 8 standard tests to aid clinicians in selecting the LOA in patients with PAD, therefore reported rates of re- 9 amputation in current practice are substantial. Up to 20% percent of above-the-knee amputations to 35% of 10 foot amputations require revision to a more proximal level. Furthermore, physician awareness of the risk for 11 re-amputation may lead to overly aggressive selection of LOA to more proximal levels in some cases. Indeed, 12 certain patients may receive amputations at a level more proximal than is necessary because their surgeon could 13 not confidently predict a high likelihood of healing at a more distal level. 14 15 In current practice, selection of LOA is determined qualitatively by clinical judgment of the surgeon using patient 16 history and physical exam. Others have developed quantitative tests that assess local microcirculation. These 17 technologies have not superseded clinical judgement. To address this critical problem, SpectralMD is 18 developing an imaging device that integrates multispectral imaging with a machine learning algorithm 19 to provide a quantitative assessment of the healing potential of a selected LOA whereas current clinical 20 practice is only capable of qualitative assessment. 21 22 We have proof-of-concept of our technology?s ability to characterize microvascular blood flow changes in a 23 patient with critical limb ischemia. In this proposal, we intend to establish the utility of our device for predicting 24 the healing potential of a clinician selected LOA by demonstrating the effectiveness of DeepView assessment 25 on a large set of pre-amputation images. To this end we will conduct a Phase I pilot study for the use of our 26 device in predicting the healing potential of a clinician selected LOA. This clinical study will develop a data set 27 for training the algorithm, and deliver a clear demonstration of feasibility for completing the development on an 28 algorithm that has high accuracy in predicting the healing potential of the proposed amputation site. Following 29 this work, we will apply for a Phase II proposal intended to finalize algorithm training, validate the algorithm 30 developed in Phase I, and establish a successful regulatory and commercialization pathway.
