% testing Deep Learning algorithms

clear all

clc

close all

addpath('functions');

% save('DL_ROI_params','stackedAEOptTheta','inputSize','hiddenSizeL1','hiddenSizeL2','outputSize','netconfig');

% save('DLconfigure/DL_LV_ES_params','stackedAEOptTheta','inputSize','hiddenSizeL1','hiddenSizeL2','outputSize','netconfig');

%%

% load images

load matFiles/validation_dataES; 

patient=3;

%min_sn=sum(slice_per_patient(1:patient-1))+1;

min_sn=1;

max_sn=sum(slice_per_patient(1:patient-1))+slice_per_patient(patient);

slice_num=min_sn:max_sn;

% get dicom info

dicom_path=['dcom/validation/ED',num2str(patient),'/images'];

para=get_dicominfo(dicom_path);

% show plots or not

disp_ena=0;

% load parameters of Deep Learning for ROI

Mroi=100;

% load Deep learning paramenters for LV segmentation

load DLconfigure/DL_LV_ES_params.mat;

for k=1:length(slice_num)

%for k=1:size(t_Iroi,3)

    C1=t_contours{k}; % read manual contours

    m_cnt=t_centers{k}; % read centers

    %subI=t_Iroi(:,:,k);

    display(['processing slice# ',num2str(slice_num(k))])

    subI=t_Iroi(:,:,slice_num(k));

    % run Deep Learning network to find the segmentation of LV

    t_yLV_h(:,:,k)=DLN(subI,stackedAEOptTheta,inputSize,hiddenSizeL1,hiddenSizeL2,outputSize,netconfig);

    % clean segmentations

    t_yLV_clean(:,:,k)=clean_segs(t_yLV_h(:,:,k));

    % dice metric

    [overlap, dice(k)]=DiceSimilarity2DImage(t_yLV_clean(:,:,k), t_yLV(:,:,k));

    % resize to original image size

    t_LVmask(:,:,k)=remap_mask(t_yLV_clean(:,:,k),t_centers{k},t_yROI(:,:,k));

    autoPoints=contourc(double(t_LVmask(:,:,k)), [0 0]);autoPoints=autoPoints(:,2:end)';

    % Perpendicular Distance

    manualPoints=t_contours{slice_num(k)};

    PD(k) = calc_dist(autoPoints,manualPoints,para);    

end

%% dice metric

avg_dice=mean(dice)

perp_dis=mean(PD)

%% display images and contours

if disp_ena==1

% get number of studies

num_studies=length(slice_per_patient);

for k=1:num_studies

    % get number of slices per patient

    spp=slice_per_patient(k);

    for k2=1:spp

        % get manual contour file name

        ind=k2+sum(slice_per_patient(1:k-1));

        figure(k)

        subplot(5,3,k2)

        imagesc(t_Iroi(:,:,ind));colormap(gray);

        hold on; contour(t_yLV_clean(:,:,ind),[0 0],'g','LineWidth',2); 

        contour(t_yLV(:,:,ind),[0 0],'r','LineWidth',2); 

        figure(num_studies+k)

        subplot(5,3,k2)

        imagesc(t_I(:,:,ind));colormap(gray);

        hold on; contour(t_LVmask(:,:,ind),[0 0],'r','LineWidth',2); 

        plot(t_contours{ind}(:,1),t_contours{ind}(:,2),'g','LineWidth',2); 

    end     

end

end

%% save contours as text files

% output=save_contours(t_LVmask,t_LV_cont_names,slice_per_patient);