Opportunity
Modern medical imaging techniques such as Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) generate high-resolution three-dimensional (3D) images that are crucial for accurate diagnosis, treatment planning, and research. However, these 3D medical images contain vast amounts of data, resulting in extremely large file sizes. This poses significant challenges for storage, transmission, and sharing within healthcare systems, particularly in telemedicine, cloud-based archiving, and collaborative research environments. Existing general-purpose image compression standards may not be optimized for the unique statistical properties and high bit-depth requirements of medical images, potentially leading to excessive loss of diagnostically critical information or inefficient compression ratios. There is a pressing need for a compression method that can significantly reduce file size while preserving the fidelity and integrity of the original 3D medical data to facilitate efficient clinical workflows and data management.
Technology
This patent presents an innovative method and system for compressing and reconstructing 3D medical images by leveraging advanced video coding technology. The core innovation lies in converting a 3D medical image into a sequence of 2D frames that can be efficiently processed by standard video codecs. The technology involves a data conversion operation that first processes the 3D image data. This typically includes partitioning the 3D volume along an axis (e.g., perpendicular to transverse, coronal, or sagittal planes) to obtain a series of 2D slices. Each pixel in these slices is mapped to an n-bit integer (e.g., 16-bit) based on the data range of the original image. A critical and innovative step is the subsequent bit-depth conversion operation, which transforms these high-bit-depth slices into video-suitable sequences (e.g., 8-bit or 10-bit). Two primary techniques are disclosed: bit-interlacing and bit-shifting. Bit-interlacing splits, for example, a 16-bit pixel into two 8-bit pixels by extracting and concatenating odd and even-position bits using bit masks, forming two separate frame sequences. Bit-shifting generates a lower bit-depth pixel (e.g., 8-bit or 10-bit) by performing a bitwise right-shift operation on the higher bits of the original pixel and discarding the lower bits. The resulting sequence of frames is then encoded using high-efficiency video coding standards like VVC, HEVC, H.264/AVC, or AVS3. These codecs exploit both inter-frame (temporal) and intra-frame (spatial) redundancies in the sequence, achieving high compression efficiency. The method also generates and encodes a data conversion indicator (flags) that records parameters like the original data range and the bit-depth conversion method used, which is essential for accurate reconstruction. For reconstruction, the process is reversed: the video bitstream is decoded, and an inverse data conversion operation (inverse bit-interlacing or inverse bit-shifting) is applied based on the decoded flags to recover the 2D slices, which are then inversely mapped and spliced back into the original 3D medical image format.
Advantages
- Achieves high compression ratios by utilizing the sophisticated redundancy reduction capabilities of modern video codecs (e.g., VVC, HEVC).
- Maintains high reconstruction fidelity, with demonstrated Peak Signal-to-Noise Ratio (PSNR) values up to 58.61 dB for CT and 63.22 dB for MRI images in tests.
- Preserves critical diagnostic information by employing controlled bit-depth conversion and mapping based on the original image's data range.
- Offers configurability and adaptability through selectable bit-depth conversion methods (bit-interlacing or bit-shifting) and support for various video coding standards.
- Enables efficient storage and faster transmission of large 3D medical datasets, benefiting telemedicine and cloud archiving.
- The system is deployable on existing video coding hardware and software platforms, facilitating integration.
- The overhead from the data conversion indicator (flags) is minimal (e.g., less than 63 bits)
Applications
- Compression and archiving of 3D medical images (e.g., MRI, CT scans) in hospital Picture Archiving and Communication Systems (PACS).
- Efficient transmission of medical images for telemedicine consultations and second opinions.
- Enabling large-scale medical image analysis and AI research by reducing dataset storage footprints.
- Facilitating real-time or rapid retrieval and visualization of 3D medical data on workstations with limited bandwidth or storage.
- Integration into medical imaging devices for on-the-fly compression before data transfer.
- Potential adaptation for compressing other types of 3D volumetric data beyond medical imaging.
