DICOM Compatibility

1. Background: PACS Integration Needs

DiCube must integrate seamlessly with existing DICOM ecosystems so PACS vendors can adopt it. DcbStreamingReader converts .dcbs files to DICOM on demand, emulating a traditional PACS backend. Downstream systems consume DiCube data without workflow changes.

2. Streaming Reader

2.1. Basic Usage

import dicube
from dicube.dicom import DcbStreamingReader

dicom_dir = 'dicube-testdata/dicom/sample_10'
dcb_file = 'dicube-testdata/sample_10.dcbs'

dcb_image = dicube.load_from_dicom_folder(dicom_dir)
dicube.save(dcb_image, dcb_file)
dcb_stream = DcbStreamingReader(dcb_file)

slice_0 = dcb_stream.get_dicom_for_frame(0)
with open('dicube-testdata/sample_10_0.dcm', 'wb') as f:
    f.write(slice_0)
print(f"Generated DICOM bytes: {len(slice_0)}")
Generated DICOM bytes: 163032

2.2. Processing + Visualization

import pydicom
import numpy as np
import matplotlib.pyplot as plt
from io import BytesIO

def analyze_dicom_slice(slice_index):
    buffer = BytesIO(dcb_stream.get_dicom_for_frame(slice_index))
    dataset = pydicom.dcmread(buffer, force=True)

    pixels = dataset.pixel_array.astype('float32')
    if hasattr(dataset, 'RescaleIntercept'):
        pixels += float(dataset.RescaleIntercept)

    plt.figure(figsize=(8, 6))
    plt.imshow(pixels, cmap='gray', vmin=-800, vmax=300)
    plt.axis('off')
    plt.title(f"Slice #{slice_index}")
    plt.tight_layout()
    plt.show()
    return dataset

for idx in [0, 4, 8]:
    analyze_dicom_slice(idx)

Conversion happens only when requested—no need to pre-expand entire series, saving space and latency.

3. HTJ2K Compatibility Landscape

3.1. Standard Evolution

DiCube uses HTJ2K (High Throughput JPEG 2000), standardized by DICOM in 2023. Compared to classic JPEG 2000:

  • 10–15× faster encoding/decoding
  • Same compression quality at equal ratios
  • Better suited to GPU/hardware acceleration

3.2. Current Support

  • Supported
    • PyDICOM ≥ 3.0.0 (with pylibjpeg-openjpeg > 2.0 + pylibjpeg > 2.0)
    • Python-GDCM 3.0.26
    • ITK-SNAP 4.4.0
  • Not yet supported
    • Horos 4.0.1 (macOS viewer)
    • SimpleITK 2.5.2

3.3. Maximum Compatibility Option

Fallback to uncompressed DICOM for legacy environments:

from dicube.dicom import DcbStreamingReader

dcb_stream_uncompressed = DcbStreamingReader(dcb_file, force_uncompressed=True)
uncompressed_dicom = dcb_stream_uncompressed.get_dicom_for_frame(0)
/home/liaofz/miniforge3/lib/python3.10/site-packages/dicube/dicom/dicom_builder.py:272: UserWarning:

Failed to decompress pixel data: 'Dataset' object has no attribute 'file_meta'

Pros: - Universally compatible - No decoder dependency - Zero risk of decode failure in transit

Cons: - 5–10× larger files - Higher storage/bandwidth usage - Longer transfer times

4. Deployment Strategy

Adopt a gradual approach:

  1. Assess HTJ2K support in target PACS/clients.
  2. Use HTJ2K where supported; fallback to uncompressed for legacy systems.
  3. Monitor upstream library updates and expand HTJ2K coverage over time.

This balances modern compression gains with real-world compatibility.