How do I correctly process PixelData to display one of the images in matplotlib?

[GOW-lover]

I work with DICONDE format files, for which pydicom is suitable with a slight modification (bytes not containing “(” are skipped)).

However, I am encountering a problem getting the images. The ds.pixel_array throws an attribute exception.
I can retrieve tags and bytes of stored images via indexes, but then when using:
dataset.BasicGrayscaleImageSequence[0].pixel_array it can't find TransferIndex.

I tried to use numpy.frombuffer, but it didn't get valid results.

I can get the PixelData, which I demonstrate in the code below. Question - how can I use this data to display in matplotlib?

ds = pydicom.dcmread(path_to_dicom_file)
print(ds.pixel_array) #Output:  AttributeError: Unable to convert the pixel data: one of Pixel Data, Float Pixel Data or Double Float Pixel Data must be present in the dataset.
print(ds.BasicGrayscaleImageSequence[0].PixelData[:15]) #Output: b'+\xaa|\xab\xad\xab\xf9\xab\xdb\xaa\xa4\xab\xfa\xaa\xf9'
arr=np.frombuffer(ds.BasicGrayscaleImageSequence[0].PixelData) 
print(arr.shape) #Output: (1283216,)

[scaramallion]

We'd really need to see an example file, but the exception message should give you a clue: all our pixel data methods depend on the dataset at the current level containing one of the pixel data elements (Pixel Data, Float Pixel Data or Double Float Pixel Data). Have you tried ds.BasicGrayscaleImageSequence[0].pixel_array?

If that doesn't work, take a look at the current pixels.decoders code for native transfer syntaxes and you should get an idea of what you need to do to get a correct numpy dtype for use with frombuffer() and how to reshape the resulting array.

[scaramallion]

From what I can tell the current DICONDE standard includes:

• E2339 - Standard Practice for Digital Imaging and Communication in Nondestructive Evaluation
• E2663 - [...] Ultrasound Methods
• E2699 - [...] Digital X-ray Methods
• E2738 - [...] Computed Radiography Methods
• E2934 - [...] Eddy Current Methods
• E2767 - [...] X-ray Computed Tomography Methods

Each priced between US$63 and US$69, although I did like the collection of 28 reference radiography TIFF files available for US$798 (RRE303001-A).

[scaramallion]

I'm actually a bit surprised out current imaging backend doesn't work if DICONDE is just a drop-in replacement for some elements... I'd really like to see an example dataset. Could you maybe post the collection of elements without their data?

Also:

Have you tried ds.BasicGrayscaleImageSequence[0].pixel_array?

That was a really stupid question, sorry about that.

And:

dataset.BasicGrayscaleImageSequence[0].pixel_array it can't find TransferIndex.

What did you mean by this? None of our code uses TransferIndex (which isn't a DICOM element). Could you post the full traceback when you try the above?

[GOW-lover]

I'm actually a bit surprised out current imaging backend doesn't work if DICONDE is just a drop-in replacement for some elements... I'd really like to see an example dataset. Could you maybe post the collection of elements without their data?

Also:

Have you tried ds.BasicGrayscaleImageSequence[0].pixel_array?

That was a really stupid question, sorry about that.

And:

dataset.BasicGrayscaleImageSequence[0].pixel_array it can't find TransferIndex.

What did you mean by this? None of our code uses TransferIndex (which isn't a DICOM element). Could you post the full traceback when you try the above?

I would have sent you an example a long time ago if I can create files of this format. I only have these files themselves and a compiled application that can open them, from where I have roughly visualized the structure of how they are contained inside.

However, this application was done by an outside organization, and I need to process the files for other tasks. I will discuss with my boss and if they give me permission, I will send you an example.

First of all. If i don't modify your lib, i receive this exception:

File "C:\Users\StrongWorkstation1\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 515, in read_sequence_item
    group, element, length = unpack(tag_length_format, bytes_read)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
struct.error: unpack requires a buffer of 8 bytes

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "f:\Downloads\DicondeLabelerActualV2\DicondeLabelerActual\DICONDE_Labeler3Edit\DICOM_TEST.py", line 17, in <module>
    ds=dcmread(path3,force=True)
       ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\ StrongWorkstation1\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 1035, in dcmread     
    dataset = read_partial(
              ^^^^^^^^^^^^^
 File "C:\Users\StrongWorkstation1\AppData\\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 857, in read_partial 
    dataset = read_dataset(
              ^^^^^^^^^^^^^
  File "C:\Users\ValeriyPC\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 427, in read_dataset 
    raw_data_element = next(de_gen)
                       ^^^^^^^^^^^^
  File "C:\Users\StrongWorkstation1\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 264, in data_element_generator
    seq = read_sequence(fp, is_implicit_VR,
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\StrongWorkstation1\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 483, in read_sequence    dataset = read_sequence_item(
              ^^^^^^^^^^^^^^^^^^^
  File "C:\Users\StrongWorkstation1\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\filereader.py", line 520, in read_sequence_item
    raise IOError(
OSError: No tag to read at file position 61EBE22

The position address varies depending on the file

if i implement some modify then i will see that:

(its bytes that throw the error above)

[b'\xfe\xff\x00\xe0\xff\xff\xff\xff'
b'\xfe\xff\x00\xe0\xff\xff\xff\xff'
b'\xfe\xff\xdd\xe0\xff\xff\xff\xff'
b'\xfe\xff\xdd\xe0\xff\xff\xff\xff'
b'\xfe\xff\xdd\xe0\xff\xff\xff\xff'
b'\xfe\xff\xdd\xe0\xff\xff\xff\xff']

C:\Users\StrongWorkstation1\AppData\Local\Programs\Python\Python312\Lib\site-packages\pydicom\valuerep.py:443: UserWarning: The value length (91) exceeds the maximum length of 64 allowed for VR UI. Please see https://dicom.nema.org/medical/dicom/current/output/html/part05.html#table_6.2-1 for allowed values for each VR.
warnings.warn(msg)

Dataset.file_meta -------------------------------
(0002, 0000) File Meta Information Group Length  UL: 140
(0002, 0001) File Meta Information Version       OB: b'\x00\x01'
(0002, 0002) Media Storage SOP Class UID         UI: Computed Radiography Image Storage
(0002, 0010) Transfer Syntax UID                 UI: Explicit VR Little Endian
(0002, 0012) Implementation Class UID            UI: 1.3.6.1.4.2.1937.1969.2
(0002, 0013) Implementation Version Name         SH: '002.289'
(0002, 0016) Source Application Entity Title     AE: '\x00\x00'
-------------------------------------------------
(0002, 0003) Media Storage SOP Instance UID      UI: 1.3.6.1.4.2.1937.1969.1.1852588488.423317911.9
(0008, 0022) Acquisition Date                    DA: '20240417'
(0008, 002a) Acquisition DateTime                DT: '20240417151830'
(0008, 0032) Acquisition Time                    TM: '151830'
(0008, 0060) Modality                            CS: 'DX'
(0008, 0201) Timezone Offset From UTC            SH: '-180'
(0014, 0028) Component Manufacturer              DS: '30.0'
(0014, 3012) Number of Frames Integrated         DS: '1.0'
(0014, 3073) Number of Frames Used for Integrati DS: '1.0'
(0018, 1050) Spatial Resolution                  DS: [130, 0]
(0018, 1150) Exposure Time                       IS: '15000'
(0018, 1164) Imager Pixel Spacing                DS: [0.124, 0.124]
(0018, 1200) Date of Last Calibration            DA: '20240417'
(0018, 1201) Time of Last Calibration            TM: '121055'
(0018, 1302) Scan Length                         IS: '317'
(0018, 7001) Detector Temperature                DS: '30.0'
(0018, 7008) Detector Mode                       LT: '5'
(0018, 700a) Detector ID                         SH: 'FDH32'
(0018, 702a) Detector Manufacturer Name          LO: 'MXD'
(0018, 9328) Exposure Time in ms                 FD: 5000.0
(0028, 0002) Samples per Pixel                   US: 1
(0028, 0004) Photometric Interpretation          CS: 'MONOCHROME1'
(0028, 0008) Number of Frames                    IS: '1'
(0028, 0010) Rows                                US: 2024
(0028, 0011) Columns                             US: 2536
(0028, 0030) Pixel Spacing                       DS: [0.124, 0.124]
(0028, 0100) Bits Allocated                      US: 16
(0028, 0101) Bits Stored                         US: 16
(0028, 0102) High Bit                            US: 15
(0028, 0103) Pixel Representation                US: 0
(0028, 1050) Window Center                       DS: '12465.0'
(0028, 1051) Window Width                        DS: '23863.0'
(0028, 2112) Lossy Image Compression Ratio       DS: '100.0'
(0070, 031a) Fiducial UID                        UI: 1.1.1.1950626654.4145252353.486086811.3496200370.1014301354.991564235.2696556857.2290699571
(2020, 0110)  Basic Grayscale Image Sequence  0 item(s) ----
(7001, 0001) Private Creator                     SL: 698
(7001, 0002) Private Creator                     OB: Array of 698 elements
(7001, 0011) Private Creator                     SL: 325
(7001, 0012) Private Creator                     OB: Array of 366 elements
(7001, 0021) Private Creator                     SL: 4
(7001, 0022)  Private Creator  0 item(s) ----
(7001, 0041) Private Creator                     SL: 86
(7001, 0042) Private Creator                     SL: 0
(7001, 0051) Private Creator                     SL: 52
(7001, 0052) Private Creator                     SL: 0
(7001, 0075) Private Creator                     SL: 0
(7001, 0076) Private Creator                     SL: 0
(7001, 0077) Private Creator                     OB: b'\x01\x00'
(7001, 0081) Private Creator                     SL: 12
(7001, 0082) Private Creator                     OB: b'\xe8\t\x00\x00\xe8\x07\x00\x00\x00\x00\x00\x00'
(7001, 0101) Private tag data                    SL: 0
(7001, 0121) Private tag data                    SL: 3
(7001, 0122)  Private tag data  0 item(s) ----
(7001, 0151) Private tag data                    SL: 72
(7001, 0152) Private tag data                    OB: Array of 72 elements
(7001, 0153) Private tag data                    SL: 56
(7001, 0154) Private tag data                    OB: Array of 56 elements
(7001, 0155) Private tag data                    SL: 13
(7001, 0156) Private tag data                    OB: b'\x00\x00\x00\x00\xe8\x10\x00\x00\x00\x00\x01\x7f\x00\x00'
(7001, 0157) Private tag data                    SL: 0
(7001, 0221) Private tag data                    OB: b'3P\xd1\xa0"\x13(G\xb9g)\x80>\x0f\xab\xd0'
(7fe0, 0010) Pixel Data                          OW: Array of 10265728 elements

As you see, in this file library didn't find series of images, but it have 10 images inside.

Since I modified your library a while ago in a slightly different way, I've forgotten what I did, but now it doesn't even see any sequences inside at all, although it did see at least the first file before.

You say that your library doesn't use TransferSyntax, but since it hasn't defined a handler in this format (BasicImageSequence), it immediately tries to get the TransferSyntaxUID from file_meta with the first line and of course it doesn't find it.
It was then throwing an exception related to this line

[scaramallion]

No, I said it doesn't use TransferIndex...

The missing Transfer Syntax UID can be filled in with:

from pydicom import FileMetaDataset

ds.BasicGrayscaleImageSequence[0].file_meta = ds.file_meta

But it doesn't look like the sequence item contains pixel data elements, so I don't think that'd be very useful anyway. But at least it should give you a new exception

[GOW-lover]

No, I said it doesn't use TransferIndex...

The missing Transfer Syntax UID can be filled in with:

from pydicom import FileMetaDataset

ds.BasicGrayscaleImageSequence[0].file_meta = ds.file_meta

But it doesn't look like the sequence item contains pixel data elements, so I don't think that'd be very useful anyway. But at least it should give you a new exception

I'll check it some later. Sequence elements has pixel data. When i tried to get metainfo by another method that i forgot (i'll try to reproduce it later) i got metainfo about first element of sequnece but it others was private. First element could be get by standard ds.pixel_array property