dicube 0.1.4__cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl

dicube/core/pixel_header.py

@@ -0,0 +1,117 @@
+import json
+from dataclasses import asdict, dataclass, field
+from typing import Dict, Optional
+
+
+@dataclass
+class PixelDataHeader:
+    """Header class for storing pixel data information in medical images.
+
+    Stores metadata including:
+    - Rescale factors (slope/intercept)
+    - Original pixel data type
+    - Window settings (center/width)
+    - Value range (min/max)
+    - Additional metadata in EXTRAS
+
+    Attributes:
+        RESCALE_SLOPE (float): Slope for linear transformation.
+        RESCALE_INTERCEPT (float): Intercept for linear transformation.
+        PIXEL_DTYPE (str): Pixel data type string (after convert to dcb file).
+        ORIGINAL_PIXEL_DTYPE (str): Original pixel data type string (before convert to dcb file).
+        WINDOW_CENTER (float, optional): Window center value for display.
+        WINDOW_WIDTH (float, optional): Window width value for display.
+        MAX_VAL (float, optional): Maximum pixel value.
+        MIN_VAL (float, optional): Minimum pixel value.
+        EXTRAS (Dict[str, any]): Dictionary for additional metadata.
+    """
+
+    RESCALE_SLOPE: float = 1.0
+    RESCALE_INTERCEPT: float = 0.0
+    ORIGINAL_PIXEL_DTYPE: str = "uint16"
+    PIXEL_DTYPE: str = "uint16"
+    WINDOW_CENTER: Optional[float] = None
+    WINDOW_WIDTH: Optional[float] = None
+    MAX_VAL: Optional[float] = None
+    MIN_VAL: Optional[float] = None
+    EXTRAS: Dict[str, any] = field(default_factory=dict)
+
+    def to_dict(self) -> dict:
+        """Convert the header to a dictionary for serialization.
+
+        Merges EXTRAS field into the main dictionary and removes
+        the redundant EXTRAS key.
+
+        Returns:
+            dict: Dictionary representation of the header.
+        """
+        data = asdict(self)
+        data.update(self.EXTRAS)  # Merge EXTRAS into dictionary
+        data.pop("EXTRAS", None)  # Remove redundant EXTRAS field
+        return data
+
+    @classmethod
+    def from_dict(cls, d: dict):
+        """Create a PixelDataHeader from a dictionary.
+
+        Args:
+            d (dict): Dictionary containing header data.
+
+        Returns:
+            PixelDataHeader: A new instance with values from the dictionary.
+        """
+        rescale_slope = d.get("RESCALE_SLOPE", 1.0)
+        rescale_intercept = d.get("RESCALE_INTERCEPT", 0.0)
+        original_pixel_dtype = d.get("ORIGINAL_PIXEL_DTYPE", "uint16")
+        window_center = d.get("WINDOW_CENTER")  # Defaults to None
+        window_width = d.get("WINDOW_WIDTH")  # Defaults to None
+        max_val = d.get("MAX_VAL")  # Defaults to None
+        min_val = d.get("MIN_VAL")  # Defaults to None
+
+        # All other keys go into EXTRAS
+        extras = {
+            k: v
+            for k, v in d.items()
+            if k
+            not in {
+                "RESCALE_SLOPE",
+                "RESCALE_INTERCEPT",
+                "ORIGINAL_PIXEL_DTYPE",
+                "WINDOW_CENTER",
+                "WINDOW_WIDTH",
+                "MAX_VAL",
+                "MIN_VAL",
+            }
+        }
+
+        return cls(
+            RESCALE_SLOPE=rescale_slope,
+            RESCALE_INTERCEPT=rescale_intercept,
+            ORIGINAL_PIXEL_DTYPE=original_pixel_dtype,
+            WINDOW_CENTER=window_center,
+            WINDOW_WIDTH=window_width,
+            MAX_VAL=max_val,
+            MIN_VAL=min_val,
+            EXTRAS=extras,
+        )
+
+    def to_json(self) -> str:
+        """Serialize the header to a JSON string.
+
+        Returns:
+            str: JSON string representation of the header.
+        """
+        return json.dumps(self.to_dict())
+
+    @classmethod
+    def from_json(cls, json_str: str):
+        """Create a PixelDataHeader from a JSON string.
+
+        Args:
+            json_str (str): JSON string containing header data.
+
+        Returns:
+            PixelDataHeader: A new instance created from the JSON data.
+        """
+        obj_dict = json.loads(json_str)
+        return cls.from_dict(obj_dict) 

dicube/dicom/__init__.py

@@ -0,0 +1,13 @@
+from .dicom_meta import DicomMeta, SortMethod, read_dicom_dir
+from .dicom_status import DicomStatus, get_dicom_status
+from .dicom_tags import CommonTags
+from .space_from_meta import get_space_from_DicomMeta
+__all__ = [
+    "DicomMeta",
+    "read_dicom_dir",
+    "DicomStatus",
+    "get_dicom_status",
+    "CommonTags",
+    "SortMethod",
+    "get_space_from_DicomMeta",
+] 

dicube/dicom/dcb_streaming.py

@@ -0,0 +1,250 @@
+"""
+DCB Streaming Reader for PACS Viewer
+
+Provides efficient streaming access to DCB files for on-demand DICOM frame delivery.
+Keeps files open and metadata cached for low-latency responses.
+"""
+
+import io
+import struct
+import warnings
+from typing import Dict, Any
+
+from pydicom import Dataset
+from pydicom.dataset import FileMetaDataset
+from pydicom.encaps import encapsulate
+from pydicom.uid import generate_uid
+import pydicom
+
+from ..storage.dcb_file import DcbFile
+from .dicom_io import save_dicom
+
+# 定义所需的最低PyDicom版本
+REQUIRED_PYDICOM_VERSION = "3.0.0"
+
+class DcbStreamingReader:
+    """
+    PACS Viewer 的 dcb 文件流式读取器
+    保持文件打开状态，支持快速随机访问帧数据
+    
+    Example:
+        reader = DcbStreamingReader('study.dcbs')
+        dicom_bytes = reader.get_dicom_for_frame(50)
+        reader.close()
+    """
+    
+    def __init__(self, dcb_file_path: str):
+        """
+        初始化并预解析所有元数据
+        
+        Args:
+            dcb_file_path: dcb 文件路径
+            
+        Warnings:
+            UserWarning: 如果 PyDicom 版本低于 3.0.0，HTJ2K 解码可能无法正常工作
+        """
+        # 检查 PyDicom 版本
+        self._check_pydicom_version()
+        
+        self.file_path = dcb_file_path
+        self.file_handle = None
+        self.transfer_syntax_uid = None
+        
+        # 预解析的数据
+        self.header = None
+        self.dicom_meta = None
+        self.pixel_header = None
+        self.space = None
+        
+        # 帧索引信息
+        self.frame_offsets = []
+        self.frame_lengths = []
+        self.frame_count = 0
+        
+        # DcbFile 实例（用于读取元数据）
+        self.dcb_file = None
+        
+        # 初始化
+        self._open_and_parse()
+    
+    def _check_pydicom_version(self):
+        """
+        检查 PyDicom 版本，如果不满足要求则发出警告
+        
+        Warnings:
+            UserWarning: 如果 PyDicom 版本低于 3.0.0
+        """
+        current_version = pydicom.__version__
+        if current_version < REQUIRED_PYDICOM_VERSION:
+            warnings.warn(
+                f"DcbStreamingReader 需要 PyDicom >= {REQUIRED_PYDICOM_VERSION} 以完全支持 HTJ2K 传输语法。"
+                f"当前 PyDicom 版本为 {current_version}，可能无法读取像素数据。"
+                f"写入功能不受影响，但其他应用读取时可能会出现问题。建议升级: pip install pydicom>={REQUIRED_PYDICOM_VERSION}，需要 python 3.10 或更高版本",
+                UserWarning
+            )
+            self._has_pydicom_htj2k_support = False
+        else:
+            self._has_pydicom_htj2k_support = True
+
+    def _open_and_parse(self):
+        """打开文件并解析所有元数据"""
+        try:
+            # 1. 创建 DcbFile 实例（会自动检测文件类型）
+            self.dcb_file = DcbFile(self.file_path, mode='r')
+            
+            # 2. 读取并缓存头部信息
+            self.header = self.dcb_file.header
+            self.frame_count = self.header['frame_count']
+            
+            # 3. 读取并缓存元数据
+            self.dicom_meta = self.dcb_file.read_meta()
+            self.pixel_header = self.dcb_file.read_pixel_header()
+            self.space = self.dcb_file.read_space()
+            
+            # 4. 获取 transfer syntax UID（从文件类型直接获取）
+            self.transfer_syntax_uid = self.dcb_file.get_transfer_syntax_uid()
+            if not self.transfer_syntax_uid:
+                # 如果文件类型没有定义 transfer syntax，使用默认的未压缩格式
+                self.transfer_syntax_uid = '1.2.840.10008.1.2.1'  # Explicit VR Little Endian
+            
+            # 5. 打开文件句柄用于读取帧数据
+            self.file_handle = open(self.file_path, 'rb')
+            
+            # 6. 读取所有帧的 offset 和 length
+            self._read_frame_indices()
+            
+        except Exception as e:
+            self.close()
+            raise RuntimeError(f"Failed to open and parse DCB file: {e}")
+    
+    def _read_frame_indices(self):
+        """读取所有帧的偏移量和长度信息"""
+        self.file_handle.seek(self.header['frame_offsets_offset'])
+        
+        # 读取 offsets
+        for _ in range(self.frame_count):
+            offset, = struct.unpack('<Q', self.file_handle.read(8))
+            self.frame_offsets.append(offset)
+        
+        # 读取 lengths
+        self.file_handle.seek(self.header['frame_lengths_offset'])
+        for _ in range(self.frame_count):
+            length, = struct.unpack('<Q', self.file_handle.read(8))
+            self.frame_lengths.append(length)
+    
+    def get_dicom_for_frame(self, frame_index: int) -> bytes:
+        """
+        获取指定帧的 DICOM 数据
+        
+        Args:
+            frame_index: 帧索引（0-based）
+            
+        Returns:
+            bytes: 完整的 DICOM 文件数据
+            
+        Raises:
+            IndexError: 如果 frame_index 超出范围
+            RuntimeError: 如果读取失败
+        """
+        # 验证索引
+        if not 0 <= frame_index < self.frame_count:
+            raise IndexError(f"Frame index {frame_index} out of range [0, {self.frame_count})")
+        
+        try:
+            # 1. 读取该帧的编码数据
+            encoded_pixel_data = self._read_encoded_frame(frame_index)
+            
+            # 2. 生成该帧的 DICOM Dataset
+            ds = self._create_dicom_dataset(frame_index, encoded_pixel_data)
+            
+            # 3. 序列化为 DICOM 文件格式
+            return self._serialize_to_dicom_bytes(ds)
+            
+        except Exception as e:
+            raise RuntimeError(f"Failed to create DICOM for frame {frame_index}: {e}")
+    
+    def _read_encoded_frame(self, frame_index: int) -> bytes:
+        """直接读取指定帧的编码数据"""
+        offset = self.frame_offsets[frame_index]
+        length = self.frame_lengths[frame_index]
+        
+        self.file_handle.seek(offset)
+        return self.file_handle.read(length)
+    
+    def _create_dicom_dataset(self, frame_index: int, encoded_data: bytes) -> Dataset:
+        """快速创建 DICOM Dataset"""
+        # 1. 从缓存的 DicomMeta 获取该帧的元数据
+        if self.dicom_meta:
+            frame_meta_dict = self.dicom_meta.index(frame_index)
+        else:
+            frame_meta_dict = {}
+        
+        # 2. 创建 Dataset
+        ds = Dataset.from_json(frame_meta_dict)
+        
+        # 3. 创建并设置文件元信息
+        file_meta = FileMetaDataset()
+        file_meta.MediaStorageSOPClassUID = ds.get('SOPClassUID', '1.2.840.10008.5.1.4.1.1.2')
+        file_meta.MediaStorageSOPInstanceUID = ds.get('SOPInstanceUID', generate_uid())
+        file_meta.TransferSyntaxUID = self.transfer_syntax_uid
+        file_meta.ImplementationClassUID = generate_uid()
+        
+        ds.file_meta = file_meta
+        
+        # 4. 确保必要的 SOP 信息
+        if not hasattr(ds, 'SOPClassUID'):
+            ds.SOPClassUID = file_meta.MediaStorageSOPClassUID
+        if not hasattr(ds, 'SOPInstanceUID'):
+            ds.SOPInstanceUID = file_meta.MediaStorageSOPInstanceUID
+        
+        # 5. 设置像素相关属性
+        if self.pixel_header:
+            ds.RescaleSlope = self.pixel_header.RESCALE_SLOPE
+            ds.RescaleIntercept = self.pixel_header.RESCALE_INTERCEPT
+        
+        # 6. 设置像素数据（使用 encapsulated format for compressed data）
+        # 压缩数据需要封装
+        ds.PixelData = encapsulate([encoded_data])
+        
+        return ds
+    
+    def _serialize_to_dicom_bytes(self, ds: Dataset) -> bytes:
+        """将 Dataset 序列化为 DICOM 文件字节流"""
+        # 使用 BytesIO 在内存中创建 DICOM 文件
+        buffer = io.BytesIO()
+        save_dicom(ds, buffer)
+        buffer.seek(0)
+        return buffer.read()
+    
+    def get_frame_count(self) -> int:
+        """获取总帧数"""
+        return self.frame_count
+    
+    def get_metadata(self) -> Dict[str, Any]:
+        """获取缓存的元数据信息"""
+        return {
+            'frame_count': self.frame_count,
+            'pixel_header': self.pixel_header.to_dict() if self.pixel_header else {},
+            'has_dicom_meta': self.dicom_meta is not None,
+            'has_space': self.space is not None,
+            'transfer_syntax': self.transfer_syntax_uid,
+            'file_type': self.dcb_file.__class__.__name__,
+        }
+    
+    def close(self):
+        """关闭文件句柄"""
+        if self.file_handle:
+            self.file_handle.close()
+            self.file_handle = None
+    
+    def __enter__(self):
+        """支持 with 语句"""
+        return self
+    
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        """退出 with 语句时自动关闭"""
+        self.close()
+    
+    def __del__(self):
+        """析构时确保文件关闭"""
+        self.close() 

dicube/dicom/dicom_io.py

@@ -0,0 +1,153 @@
+import os
+import warnings
+import inspect
+from typing import List, Optional
+
+import numpy as np
+from pydicom import Dataset
+from pydicom.dataset import FileMetaDataset
+from pydicom.uid import JPEG2000, ExplicitVRLittleEndian, JPEG2000Lossless, generate_uid
+
+from ..dicom.dicom_meta import DicomMeta
+
+
+def prepare_output_dir(output_dir: str):
+    """Prepare output directory"""
+    if os.path.exists(output_dir):
+        import shutil
+
+        shutil.rmtree(output_dir)
+    os.makedirs(output_dir, exist_ok=True)
+
+
+def create_file_meta(ds):
+    """Create file meta information"""
+    file_meta = FileMetaDataset()
+
+    MODALITY_SOP_CLASS_MAP = {
+        "CT": "1.2.840.10008.5.1.4.1.1.2",
+        "MR": "1.2.840.10008.5.1.4.1.1.4",
+        "US": "1.2.840.10008.5.1.4.1.1.6.1",
+        "PT": "1.2.840.10008.5.1.4.1.1.128",
+        "CR": "1.2.840.10008.5.1.4.1.1.1",
+        "DX": "1.2.840.10008.5.1.4.1.1.1.1",
+        "NM": "1.2.840.10008.5.1.4.1.1.20",
+    }
+
+    modality = ds.Modality if hasattr(ds, "Modality") else "CT"
+    default_sop_uid = MODALITY_SOP_CLASS_MAP.get(modality, MODALITY_SOP_CLASS_MAP["CT"])
+
+    file_meta.MediaStorageSOPClassUID = (
+        ds.SOPClassUID if hasattr(ds, "SOPClassUID") else default_sop_uid
+    )
+    file_meta.MediaStorageSOPInstanceUID = (
+        ds.SOPInstanceUID if hasattr(ds, "SOPInstanceUID") else generate_uid()
+    )
+    file_meta.TransferSyntaxUID = ExplicitVRLittleEndian
+    file_meta.ImplementationClassUID = generate_uid()
+
+    return file_meta
+
+
+def ensure_required_tags(ds):
+    """Ensure required DICOM tags exist"""
+    if not hasattr(ds, "SOPClassUID"):
+        ds.SOPClassUID = ds.file_meta.MediaStorageSOPClassUID
+    if not hasattr(ds, "SOPInstanceUID"):
+        ds.SOPInstanceUID = ds.file_meta.MediaStorageSOPInstanceUID
+
+
+def set_dicom_pixel_attributes(img, ds):
+    """Set DICOM pixel attributes"""
+    if np.issubdtype(img.dtype, np.integer):
+        bits = img.dtype.itemsize * 8
+        ds.BitsAllocated = bits
+        ds.BitsStored = bits
+        ds.HighBit = bits - 1
+        ds.PixelRepresentation = 1 if np.issubdtype(img.dtype, np.signedinteger) else 0
+    else:
+        warnings.warn(f"Converting float dtype {img.dtype} to uint16")
+        img = img.astype(np.uint16)
+        ds.BitsAllocated = 16
+        ds.BitsStored = 16
+        ds.HighBit = 15
+        ds.PixelRepresentation = 0
+
+    ds.SamplesPerPixel = 1
+    return img
+
+
+def create_dicom_dataset(meta_dict: dict, pixel_header):
+    """Create DICOM dataset"""
+    ds = Dataset.from_json(meta_dict)
+
+    if hasattr(ds, "file_meta"):
+        warnings.warn("Found original file metadata, will be overridden")
+
+    ds.file_meta = create_file_meta(ds)
+    ensure_required_tags(ds)
+    ds.RescaleSlope = pixel_header.RESCALE_SLOPE
+    ds.RescaleIntercept = pixel_header.RESCALE_INTERCEPT
+
+    return ds
+
+
+def save_dicom(ds: Dataset, output_path: str):
+    """保存DICOM文件
+    
+    参数:
+        ds: DICOM数据集
+        output_path: 输出文件路径
+    """
+    
+    sig = inspect.signature(Dataset.save_as)
+    if "enforce_file_format" in sig.parameters:  # pydicom >= 3.0
+        ds.save_as(output_path, enforce_file_format=True)
+    else:
+        # 确保使用有效的传输语法UID
+        if hasattr(ds, 'file_meta') and hasattr(ds.file_meta, 'TransferSyntaxUID'):
+            # 检查是否有效，如果无效则替换为标准的ExplicitVRLittleEndian
+            try:
+                from pydicom.uid import UID
+                uid = UID(ds.file_meta.TransferSyntaxUID)
+                if not hasattr(uid, 'is_little_endian'):
+                    ds.file_meta.TransferSyntaxUID = ExplicitVRLittleEndian
+            except (ValueError, AttributeError):
+                # 如果UID无效，使用标准的ExplicitVRLittleEndian
+                ds.file_meta.TransferSyntaxUID = ExplicitVRLittleEndian
+        ds.save_as(output_path, write_like_original=False)
+
+def save_to_dicom_folder(
+    raw_images: np.ndarray,
+    dicom_meta: DicomMeta,
+    pixel_header,
+    output_dir: str,
+    filenames: Optional[List[str]] = None,
+):
+    """将图像数据保存为DICOM文件夹
+    
+    参数:
+        raw_images: 原始图像数据，可以是2D或3D数组
+        dicom_meta: DICOM元数据
+        pixel_header: 像素头信息
+        output_dir: 输出目录
+        filenames: 自定义文件名列表，如果为None则使用默认名称
+    """
+    prepare_output_dir(output_dir)
+
+    if raw_images.ndim == 2:
+        raw_images = raw_images[np.newaxis, ...]
+
+    for idx in range(len(raw_images)):
+        ds = create_dicom_dataset(dicom_meta.index(idx), pixel_header)
+        img = raw_images[idx]
+
+        if img.dtype != np.uint16:
+            img = set_dicom_pixel_attributes(img, ds)
+
+        ds.PixelData = img.tobytes()
+
+        output_path = os.path.join(
+            output_dir, filenames[idx] if filenames else f"slice_{idx:04d}.dcm"
+        )
+        save_dicom(ds, output_path)