Table 4 Pseudo code of the proposed methodology to find CV of ECG localized features.

1. Requirement: Find the CV distribution of all the localized features

2. E(t) = Healthy + Unhealthy ECG signal

3. E_f(t) = filtering(E(t))

4. Normalization: E_n(t) = (E_f(t) − E_fmin(t))/(E_fmax(t) − E_fmin(t))

5. Extraction of boundaries:

    for i = 1 to length (E_n(t))

                         {E(1), E(2), E(3), —E(n),} = BD[E_n(t)]

    endfor

    where E(n) is the nth ECG beat.

6. Extracting ECG Features:

    for i = 1 to n

                         {PR[i], QRS[i], QT[i]} = FE[E(i)]

    endfor

7. Taking QT interval array for PSR analysis:

    QT(i), i ∈ {1, 2, 3, ....n}

8. Applying Sliding window on QT interval array:

    for z = 1 to n − 19

    SW(t) = {QT(z + 1), QT(z + 2), QT(z + 3)....QT(z + 19)}

    PSR_Image(k) = plot(SW(t), SW(t − 20))

    endfor

    SW = Sliding Window.

9. Box-count array:

    for x = 1 to k

    Box_count_array(x) = No. ofBlackBoxes(PSR_image(x))

    endfor

10. Sliding window on Box-count distribution:

       for m = 1 to x − 19

       CV(m) = CV(Box_count_array(m), Box_count_array(2), Box_count_array(3))...Box_count_array(m + 19)

       endfor

11. for i = 1 to m

       Th_max(i) = max[CV(i){1:a}]

       Th_min1(i) = min[CV(i){b:end}]

       endfor Note: point ‘a’ and ‘b’ in CV plot are corresponding to ‘Unhealhty QT start’ and ‘Healthy QT end’ windows of Fig. 14

12. Th_{final_max_QT} = max(Th_max(1), Th_max(2), Th_max(3)..Th_max(m))

       Th_{final_min_QT} = min(Th_min(1), Th_min(2), Th_min(3)..Th_min(m))

13. Th_{final_min_QT} and Th_{final_max_QT} are the two thresholds of CV corresponing to QT interval, the similar procedure is followed for PR interval and QRS complex to find thresholds.