How to convert DICOM-RT structure Contour data into image coordinate?

Question

I use Python-pydicom module to load a patient sample:

def load_data(sample):
    data=dict()
    # dcms data
    dcms_data=dict()
    for dcm_file in sample['dcm_files']:  # 遍历读取数据
        ds = pydicom.dcmread(dcm_file)
        array = ds.pixel_array
        origin = ds.ImagePositionPatient  # 网格原点在世界坐标系的位置
        spacing = ds.PixelSpacing  # 采样间隔
        uid = ds.SOPInstanceUID
        dcms_data[uid] = {'dcmSpacing': spacing, 'dcmOrigin': origin, 'array': array}
    data['dcms_data']=dcms_data

    # rt data
    rt_data = dict()  # 以{RUID:label_data}形式返回结果
    ds = pydicom.dcmread(sample['rt_file'])
    sequences = ds.ROIContourSequence[0].ContourSequence
    for sequence in sequences:
        ruid = sequence.ContourImageSequence[0].ReferencedSOPInstanceUID
        array = sequence.ContourData
        num = sequence.NumberOfContourPoints
        rt_data[ruid] = {'pointNumber': num, 'array': array}
    data['rt_data']=rt_data
    # 返回结果
    return data

Then I convert the DICOM-RT struct contour data into image coordinate:

def convert_global_aix_to_net_pos(data):
    point_data = {}  # 返回坐标{uid:data}
    for uid, value in data['rt_data'].items():
        num = value['pointNumber']
        label_data = value['array']
        dcm_origin = data['dcms_data'][uid]['dcmOrigin']
        dcm_spacing = data['dcms_data'][uid]['dcmSpacing']

        point = []  # 坐标[(x1,y1),(...),...]
        for i in range(0,num,3):
            x = label_data[i]  # 轮廓世界坐标系
            y = label_data[i + 1]
            X = int(float(x) - float(dcm_origin[0]) / float(dcm_spacing[0]))  # 轮廓X坐标
            Y = int(float(y) - float(dcm_origin[1]) / float(dcm_spacing[1]))  # 轮廓Y坐标
            point.append((X, Y))
        point_data[uid] = point
    return point_data

However when I test this function on my dicom files.I find that it returned wrong points data(negative data)

I guess the method I convert DICOM-RT Struct contour data into image coordinate is wrong,but I just can't find another way. Is my method wrong? Or How could I implement it? Thanks in advance.

Answer 1

It looks like you are missing brackets around the float(x) - float(dcm_origin[0]) (and similarly for the y line). The subtraction needs to be done before the division.

Otherwise it looks okay, assuming ImageOrientationPatient of (1, 0, 0, 0, 1, 0).