Segmentation fault with odd element length

[xf4j]

When running "dcm2niix ~/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261", it shows an error with the following log. I tried with different dcm2niix arguments but the error was the same.

Compression will be faster with 'pigz' installed http://macappstore.org/pigz/
Chris Rorden's dcm2niiX version v1.0.20231111 (JP2:OpenJPEG) (JP-LS:CharLS) Clang15.0.0 x86-64 (64-bit MacOS)
Found 276 DICOM file(s)
Illegal DICOM tag 0069,005b (odd element length -4587521): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477593.597542.dcm
Illegal DICOM tag ffda,ffca (odd element length -3080285): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477601.475800.dcm
Illegal DICOM tag fff0,ffdd (odd element length 2359355): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477595.186512.dcm
Assuming final tag is Pixel Data (7fe0,0010) (issue 639)
Assuming final tag is Pixel Data (7fe0,0010) (issue 639)
Assuming final tag is Pixel Data (7fe0,0010) (issue 639)
Illegal DICOM tag fc89,fc6b (odd element length -63964073): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477581.401854.dcm
Assuming final tag is Pixel Data (7fe0,0010) (issue 639)
Illegal DICOM tag ffa8,ffba (odd element length -14090429): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477601.475791.dcm
Illegal DICOM tag fc44,fc36 (odd element length -62718873): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477577.894328.dcm
Illegal DICOM tag 001c,ffdd (odd element length -4718723): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477595.186507.dcm
Illegal DICOM tag fc6b,fc9b (odd element length -59114383): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477586.747558.dcm
Illegal DICOM tag 0042,004d (odd element length -3407915): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477593.597541.dcm
Illegal DICOM tag ffdd,0013 (odd element length 393195): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477595.186505.dcm
Illegal DICOM tag ff81,ff76 (odd element length -1245247): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477601.867695.dcm
Illegal DICOM tag ff93,ff5f (odd element length -4849723): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477601.837115.dcm
Illegal DICOM tag ffa6,ffa8 (odd element length -12058721): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477601.475793.dcm
Illegal DICOM tag ffcb,ffaa (odd element length -32178367): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477588.248646.dcm
Illegal DICOM tag fc5d,fc85 (odd element length -59114383): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477581.401842.dcm
Illegal DICOM tag 0052,0040 (odd element length 8912967): /Users/xuefeng/Downloads/1.2.392.200036.9116.2.6.1.37.2430222242.1668477554.629261/1.2.392.200036.9116.2.6.1.37.2430222242.1668477576.168972.dcm
Assuming final tag is Pixel Data (7fe0,0010) (issue 639)
Segmentation fault: 11

Version:
Chris Rorden's dcm2niiX version v1.0.20231111 (JP2:OpenJPEG) (JP-LS:CharLS) Clang15.0.0 x86-64 (64-bit MacOS)
v1.0.20231111

Error happens on both MacOS and Linux (Ubuntu 20.04), same dcm2niix version.

This DICOM series can be downloaded from https://drive.google.com/file/d/10D85yRRp18_QIdf93l3n3kaC0mNUggQ1/view?usp=sharing

Thank you very much!

[neurolabusc]

This is a limitation with your DICOM images, not with dcm2niix. You can verify this by using another tool like dcmdump:

dcmdump 1.2.392.200036.9116.2.6.1.37.2430222242.1668477601.867704.dcm > 704.txt  
W: Found element (0029,e131) with VR UN and undefined length, reading a sequence with transfer syntax LittleEndianImplicit (CP-246)

The issue is that an anonymization tool was not aware that the private tag 2005,140f is a Philips Private SQ. Perhaps the input image was saved as implicit VRs which can obscure this fact for tools that do not contain a complete DICOM dictionary.

I can think of two solutions:

1. Check the provenance of the DICOM images, and convert the original images prior to the data corruption. I would strongly prefer this option, as this will have access to additional tags that have been anonymized.
2. Clone the dcm2niix repository and change one line of code, and recompile dcm2niix. This will now convert your corrupted images, but it may feel with original Philips data and the corrupted files may be missing other details.

If you want to pursue option 2, you would change:

uint32_t array[array_size] = {0x2005 + (uint32_t(0x140F) << 16), 0x0008 + (uint32_t(0x1111) << 16), 0x0008 + (uint32_t(0x1115) << 16), 0x0008 + (uint32_t(0x1140) << 16), 0x0008 + (uint32_t(0x1199) << 16), 0x0008 + (uint32_t(0x2218) << 16), 0x0008 + (uint32_t(0x9092) << 16), 0x0018 + (uint32_t(0x9006) << 16), 0x0018 + (uint32_t(0x9042) << 16), 0x0018 + (uint32_t(0x9045) << 16), 0x0018 + (uint32_t(0x9049) << 16), 0x0018 + (uint32_t(0x9112) << 16), 0x0018 + (uint32_t(0x9114) << 16), 0x0018 + (uint32_t(0x9115) << 16), 0x0018 + (uint32_t(0x9117) << 16), 0x0018 + (uint32_t(0x9119) << 16), 0x0018 + (uint32_t(0x9125) << 16), 0x0018 + (uint32_t(0x9152) << 16), 0x0018 + (uint32_t(0x9176) << 16), 0x0018 + (uint32_t(0x9226) << 16), 0x0018 + (uint32_t(0x9239) << 16), 0x0020 + (uint32_t(0x9071) << 16), 0x0020 + (uint32_t(0x9111) << 16), 0x0020 + (uint32_t(0x9113) << 16), 0x0020 + (uint32_t(0x9116) << 16), 0x0020 + (uint32_t(0x9221) << 16), 0x0020 + (uint32_t(0x9222) << 16), 0x0028 + (uint32_t(0x9110) << 16), 0x0028 + (uint32_t(0x9132) << 16), 0x0028 + (uint32_t(0x9145) << 16), 0x0040 + (uint32_t(0x0260) << 16), 0x0040 + (uint32_t(0x0555) << 16), 0x0040 + (uint32_t(0xa170) << 16), 0x5200 + (uint32_t(0x9229) << 16), 0x5200 + (uint32_t(0x9230) << 16)};

to read:

uint32_t array[array_size] = {0x0008 + (uint32_t(0x1111) << 16), 0x0008 + (uint32_t(0x1115) << 16), 0x0008 + (uint32_t(0x1140) << 16), 0x0008 + (uint32_t(0x1199) << 16), 0x0008 + (uint32_t(0x2218) << 16), 0x0008 + (uint32_t(0x9092) << 16), 0x0018 + (uint32_t(0x9006) << 16), 0x0018 + (uint32_t(0x9042) << 16), 0x0018 + (uint32_t(0x9045) << 16), 0x0018 + (uint32_t(0x9049) << 16), 0x0018 + (uint32_t(0x9112) << 16), 0x0018 + (uint32_t(0x9114) << 16), 0x0018 + (uint32_t(0x9115) << 16), 0x0018 + (uint32_t(0x9117) << 16), 0x0018 + (uint32_t(0x9119) << 16), 0x0018 + (uint32_t(0x9125) << 16), 0x0018 + (uint32_t(0x9152) << 16), 0x0018 + (uint32_t(0x9176) << 16), 0x0018 + (uint32_t(0x9226) << 16), 0x0018 + (uint32_t(0x9239) << 16), 0x0020 + (uint32_t(0x9071) << 16), 0x0020 + (uint32_t(0x9111) << 16), 0x0020 + (uint32_t(0x9113) << 16), 0x0020 + (uint32_t(0x9116) << 16), 0x0020 + (uint32_t(0x9221) << 16), 0x0020 + (uint32_t(0x9222) << 16), 0x0028 + (uint32_t(0x9110) << 16), 0x0028 + (uint32_t(0x9132) << 16), 0x0028 + (uint32_t(0x9145) << 16), 0x0040 + (uint32_t(0x0260) << 16), 0x0040 + (uint32_t(0x0555) << 16), 0x0040 + (uint32_t(0xa170) << 16), 0x5200 + (uint32_t(0x9229) << 16), 0x5200 + (uint32_t(0x9230) << 16)};

[neurolabusc]

I have made a commit to the development branch so it will convert your images. This is based on the fact that your data is from Canon (nee Toshiba) rather than Philips. It does seem like the data includes a lot of stripped and retired tags, so I do still encourage you to ensure conversion with a image direct from the instrument.

[xf4j]

Thank you.