Measuring the reliability of a HSMM heart sound segmentation algorithm
; Mantadelis, T.M
; Gomes, P.
Measuring the reliability of a HSMM heart sound segmentation algorithm, Proc Portuguese Conf. on Pattern Recognition - RecPad, Porto, Portugal, Vol. , pp. - , October, 2019.
Digital Object Identifier:
Download Full text PDF ( 282 KBs)
In this paper, we use a conditional distribution in order to compute the
confidence of our segmentation. Using the proposed metric noisy samples
are identified and removed from the recording, and as a result, trustful
samples are leaved for a further post-processing. By using the proposed
confidence metric, we were able to identify wrong classification samples
and boost our system (in average) from an ≈ 83% up to ≈ 90% of positive
predictability per sample.