Options handling improvements.

app.py

@@ -145,7 +145,7 @@ def predict(numberOfECGs:         int = 1,
       deepfakeecg.generateDeepfakeECGs(numberOfECGs,
                                        ecgType            = ecgType,
                                        ecgLengthInSeconds = ecgLengthInSeconds,
-                                       ecgScaleFactor     = 6,
+                                       ecgScaleFactor     = deepfakeecg.ECG_DEFAULT_SCALE_FACTOR,
                                        outputFormat       = deepfakeecg.OUTPUT_TENSOR,
                                        showProgress       = False,
                                        runOnDevice        = runOnDevice)
@@ -157,7 +157,10 @@ def predict(numberOfECGs:         int = 1,
    for result in Sessions[request.session_hash].Results:
 
       # ====== Plot ECG =====================================================
-      result = result.t().detach().cpu().numpy()
+      # 1. Convert to NumPy
+      # 2. Remove the Timestamp column (0)
+      # 3. Convert from µV to mV
+      result = result.t().detach().cpu().numpy()[1:] / 1000
       # print(result)
 
       # ------ ECG-12 -------------------------------------------------------
@@ -199,51 +202,6 @@ def select(event:   gradio.SelectData,
    log(f'Session "{request.session_hash}": Selected ECG #{Sessions[request.session_hash].Selected + 1}')
 
 
-# ###### Produce ECG CSV file from Tensor ###################################
-def dataToCSV(ecgResult, ecgType, outputFileName):
-
-   data = ecgResult.detach().cpu().numpy()
-
-   if ecgType == deepfakeecg.DATA_ECG8:
-      header = 'Timestamp,LeadI,LeadII,V1,V2,V3,V4,V5,V6'
-   elif ecgType == deepfakeecg.DATA_ECG12:
-      header = 'Timestamp,LeadI,LeadII,V1,V2,V3,V4,V5,V6,LeadIII,aVL,aVR,aVF'
-   else:
-      raise Exception('Invalid ECG type!')
-
-   numpy.savetxt(outputFileName, data,
-                 header    = header,
-                 comments  = '',
-                 delimiter = ',',
-                 fmt       = '%i')
-
-
-# ###### Produce ECG PDF file from Tensor ###################################
-def dataToPDF(ecgResult, ecgType, outputFileName):
-
-   data = ecgResult.detach().cpu().numpy()
-   outputLeads = deepfakeecg.ECG_LEADS
-
-   matplotlib.pyplot.figure(figsize=(15, 3))
-   for outputLead in outputLeads:
-      try:
-         outputLeadIndex = deepfakeecg.ECG_LEADS[outputLead][0]
-         outputLeadLabel = deepfakeecg.ECG_LEADS[outputLead][1]
-         outputLeadType  = deepfakeecg.ECG_LEADS[outputLead][2]
-      except:
-         raise Exception('Invalid lead ' + outputLead + '!')
-      if outputLeadType > ecgType:
-         raise Exception('Invalid lead ' + outputLead + ' for this ECG type!')
-      matplotlib.pyplot.plot(data[:, outputLeadIndex], label = outputLeadLabel)
-   matplotlib.pyplot.legend()
-   matplotlib.pyplot.title('Generated ECG — ID ' + str(i))
-   matplotlib.pyplot.xlabel('Time [s]')
-   matplotlib.pyplot.ylabel('Amplitude [μV]')
-   matplotlib.pyplot.grid(True)
-   matplotlib.pyplot.ylim(-1000, +1000)
-   matplotlib.pyplot.savefig(outputFileName)
-
-
 # ###### Download CSV #######################################################
 def downloadCSV(request: gradio.Request) -> pathlib.Path:
 
@@ -251,7 +209,7 @@ def downloadCSV(request: gradio.Request) -> pathlib.Path:
    ecgType   = Sessions[request.session_hash].Type
    fileName  = pathlib.Path(Sessions[request.session_hash].TempDirectory.name) / \
                   ('ECG-' + str(Sessions[request.session_hash].Selected + 1) + '.csv')
-   dataToCSV(ecgResult, ecgType, fileName)
+   deepfakeecg.dataToCSV(ecgResult, ecgType, fileName)
 
    log(f'Session "{request.session_hash}": Download CSV file {fileName}')
    return fileName
@@ -265,7 +223,13 @@ def downloadPDF(request: gradio.Request):
    ecgType   = Sessions[request.session_hash].Type
    fileName  = pathlib.Path(Sessions[request.session_hash].TempDirectory.name) / \
                   ('ECG-' + str(Sessions[request.session_hash].Selected + 1) + '.pdf')
-   dataToPDF(ecgResult, ecgType, fileName)
+   if ecgType == deepfakeecg.DATA_ECG12:
+      outputLeads = [ 'I', 'II', 'III', 'aVL', 'aVR', 'aVF', 'V1', 'V2', 'V3', 'V4' , 'V5' , 'V6' ]
+   else:
+      outputLeads = [ 'I', 'II', 'V1', 'V2', 'V3', 'V4' , 'V5' , 'V6' ]
+
+   deepfakeecg.dataToPDF(ecgResult, ecgType, outputLeads, fileName,
+                         Sessions[request.session_hash].Selected + 1)
 
    log(f'Session "{request.session_hash}": Download PDF file {fileName}')
    return fileName

requirements.txt

@@ -4,4 +4,4 @@ matplotlib
 Pillow
 pydantic==2.10.6
 torch
-git+https://github.com/dreibh/deepfake-ecg@main#egg=deepfake_ecg
+git+https://github.com/dreibh/deepfake-ecg@dreibh/export-improvements#egg=deepfake_ecg