PET/CT parsing to a Numpy 3D array

PET_parsing.py

import dicom
import os
import numpy
import pylab

# readind data
ds = dicom.read_file("DICOMDIR") # read a dicomdir of PET/CT

# collect pixel_volume and resolution

def getSec(s):
    b =int(s[0:2]) *3600 + int(s[2:4])*60 + int(s[4:6])
    return b

for record in ds.DirectoryRecordSequence:
    if record.DirectoryRecordType == "IMAGE":
        if record.ImageID == "PET AC 2D":
            path = os.path.join(*record.ReferencedFileID)
            dcm = dicom.read_file(path)
            pixel_spacing = dcm.PixelSpacing[1]
            slice_thickness = record.SliceThickness
            row = dcm.Rows
            column = dcm.Columns
            slice_number = dcm.InstanceNumber
            body_weight = dcm.PatientWeight *1000
            injection_time = getSec(dcm.RadiopharmaceuticalInformationSequence[0].RadiopharmaceuticalStartTime)
            acqusition_time = getSec(dcm.AcquisitionTime)
            dose = dcm.RadiopharmaceuticalInformationSequence[0].RadionuclideTotalDose
            half_life = dcm.RadiopharmaceuticalInformationSequence[0].RadionuclideHalfLife
            break
actural_dose = dose * 0.5**((acqusition_time-injection_time)/half_life)

pixel_volume = pixel_spacing ** 2 * slice_thickness/1000
print("pixel size =", pixel_spacing, "x", pixel_spacing, "x", slice_thickness, "mm")
print("pixel volume=", pixel_volume, "cc")
print("This image is", row, "x", column, "x", slice_number)

pixel_data = numpy.ones((slice_number,row,column))

#combine the arrays into 3D dataset
i = slice_number-1 #Because our scan is from pelvis to head
for record in ds.DirectoryRecordSequence: #https://groups.google.com/forum/#!msg/pydicom/TR2nl6_JRtM/aDODx-eNUTcJ
    if record.DirectoryRecordType == "IMAGE":
        if record.ImageID == "PET AC 2D":
            # Extract the relative path to the DICOM file
            path = os.path.join(*record.ReferencedFileID)
            dcm = dicom.read_file(path)
            SUV = dcm.pixel_array * body_weight / actural_dose * record[0x0028, 0x1053].value # Rescale Slope
            # Now get your image data
            pixel_data[i,:,:] = SUV 
            i = i-1

Your conversion into seconds is false
def getSec(s):
b =int(s[0:2]) *3600 + int(s[2:4])*60 + int(s[5:6]).
return b

It should be :
def getSec(s):
b =int(s[0:2]) *3600 + int(s[2:4])*60 + int(s[4:6]).
return b

Also be carefull as AcquisitionTime change over slices

Thanks a lot