class ImageFileReader(object): """Reader for DICOM datasets representing Image Information Entities. It provides efficient access to individual Frame items contained in the Pixel Data element without loading the entire element into memory. Attributes ---------- filename: str Path to the DICOM Part10 file on disk Examples -------- >>> from highdicom.io import ImageFileReader >>> with ImageFileReader('/path/to/file.dcm') as image: ... print(image.metadata) ... for i in range(image.number_of_frames): ... frame = image.read_frame(i) ... print(frame.shape) """ def __init__(self, filename: str): """ Parameters ---------- filename: str Path to a DICOM Part10 file containing a dataset of an image SOP Instance """ self.filename = filename def __enter__(self) -> 'ImageFileReader': self.open() return self def __exit__(self, except_type, except_value, except_trace) -> None: self._fp.close() if except_value: sys.stdout.write('Error while accessing file "{}":\n{}'.format( self.filename, str(except_value))) for tb in traceback.format_tb(except_trace): sys.stdout.write(tb) raise def open(self) -> None: """Opens file and reads metadata from it. Raises ------ FileNotFoundError When file cannot be found OSError When file cannot be opened IOError When DICOM metadata cannot be read from file ValueError When DICOM dataset contained in file does not represent an image Note ---- Builds a Basic Offset Table to speed up subsequent frame-level access. """ logger.debug('read File Meta Information') file_meta = read_file_meta_info(self.filename) transfer_syntax_uid = UID(file_meta.TransferSyntaxUID) try: self._fp = DicomFile(str(self.filename), mode='rb') self._fp.is_little_endian = transfer_syntax_uid.is_little_endian self._fp.is_implicit_VR = transfer_syntax_uid.is_implicit_VR except FileNotFoundError: raise FileNotFoundError(f'File not found: "{self.filename}"') except Exception: raise OSError( f'Could not open file for reading: "{self.filename}"') logger.debug('read metadata elements') try: self._metadata = dcmread(self._fp, stop_before_pixels=True) except Exception as err: raise IOError( f'DICOM metadata cannot be read from file "{self.filename}": ' f'"{err}"') self._pixel_data_offset = self._fp.tell() # Determine whether dataset contains a Pixel Data element try: tag = TupleTag(self._fp.read_tag()) except EOFError: raise ValueError( 'Dataset does not represent an image information entity.') if int(tag) not in _PIXEL_DATA_TAGS: raise ValueError( 'Dataset does not represent an image information entity.') self._as_float = False if int(tag) in _FLOAT_PIXEL_DATA_TAGS: self._as_float = True # Reset the file pointer to the beginning of the Pixel Data element self._fp.seek(self._pixel_data_offset, 0) # Build the ICC Transformation object. This takes some time and should # be done only once to speedup subsequent color corrections. if self.metadata.SamplesPerPixel == 1: self._color_manager = None else: try: icc_profile = self.metadata.ICCProfile except AttributeError: try: if len(self.metadata.OpticalPathSequence) > 1: # This should not happen in case of a color image. logger.warning( 'color image contains more than one optical path') optical_path_item = self.metadata.OpticalPathSequence[0] icc_profile = optical_path_item.ICCProfile except (IndexError, AttributeError): raise AttributeError( 'No ICC Profile found in image metadata.') try: self._color_manager = ColorManager(icc_profile) except ValueError: logger.warning('could not read ICC Profile') self._color_manager = None logger.debug('build Basic Offset Table') transfer_syntax_uid = self.metadata.file_meta.TransferSyntaxUID if transfer_syntax_uid.is_encapsulated: try: self._basic_offset_table = _get_bot( self._fp, number_of_frames=self.number_of_frames) except Exception as err: raise IOError(f'Failed to build Basic Offset Table: "{err}"') self._first_frame_offset = self._fp.tell() else: if self._fp.is_implicit_VR: header_offset = 4 + 4 # tag and length else: header_offset = 4 + 2 + 2 + 4 # tag, VR, reserved and length self._first_frame_offset = self._pixel_data_offset + header_offset n_pixels = self._pixels_per_frame bits_allocated = self.metadata.BitsAllocated if bits_allocated == 1: self._basic_offset_table = [ int(np.floor(i * n_pixels / 8)) for i in range(self.number_of_frames) ] else: self._basic_offset_table = [ i * self._bytes_per_frame_uncompressed for i in range(self.number_of_frames) ] if len(self._basic_offset_table) != self.number_of_frames: raise ValueError( 'Length of Basic Offset Table does not match Number of Frames.' ) @property def metadata(self) -> Dataset: """pydicom.dataset.Dataset: Metadata""" try: return self._metadata except AttributeError: raise IOError('File has not been opened for reading.') @property def _pixels_per_frame(self) -> int: """int: Number of pixels per frame""" return int( np.prod([ self.metadata.Rows, self.metadata.Columns, self.metadata.SamplesPerPixel ])) @property def _bytes_per_frame_uncompressed(self) -> int: """int: Number of bytes per frame when uncompressed""" n_pixels = self._pixels_per_frame bits_allocated = self.metadata.BitsAllocated if bits_allocated == 1: # Determine the nearest whole number of bytes needed to contain # 1-bit pixel data. e.g. 10 x 10 1-bit pixels is 100 bits, which # are packed into 12.5 -> 13 bytes return n_pixels // 8 + (n_pixels % 8 > 0) else: return n_pixels * bits_allocated // 8 def close(self) -> None: """Closes file.""" self._fp.close() def read_frame_raw(self, index: int) -> bytes: """Reads the raw pixel data of an individual frame item. Parameters ---------- index: int Zero-based frame index Returns ------- bytes Pixel data of a given frame item encoded in the transfer syntax. Raises ------ IOError When frame could not be read """ if index > self.number_of_frames: raise ValueError('Frame index exceeds number of frames in image.') logger.debug(f'read frame #{index}') frame_offset = self._basic_offset_table[index] self._fp.seek(self._first_frame_offset + frame_offset, 0) if self.metadata.file_meta.TransferSyntaxUID.is_encapsulated: try: stop_at = self._basic_offset_table[index + 1] - frame_offset except IndexError: # For the last frame, there is no next offset available. stop_at = -1 n = 0 # A frame may consist of multiple items (fragments). fragments = [] while True: tag = TupleTag(self._fp.read_tag()) if n == stop_at or int(tag) == SequenceDelimiterTag: break if int(tag) != ItemTag: raise ValueError(f'Failed to read frame #{index}.') length = self._fp.read_UL() fragments.append(self._fp.read(length)) n += 4 + 4 + length frame_data = b''.join(fragments) else: frame_data = self._fp.read(self._bytes_per_frame_uncompressed) if len(frame_data) == 0: raise IOError(f'Failed to read frame #{index}.') return frame_data def read_frame(self, index: int, correct_color: bool = True) -> np.ndarray: """Reads and decodes the pixel data of an individual frame item. Parameters ---------- index: int Zero-based frame index correct_color: bool, optional Whether colors should be corrected by applying an ICC transformation. Will only be performed if metadata contain an ICC Profile. Returns ------- numpy.ndarray Array of decoded pixels of the frame with shape (Rows x Columns) in case of a monochrome image or (Rows x Columns x SamplesPerPixel) in case of a color image. Raises ------ IOError When frame could not be read """ frame_data = self.read_frame_raw(index) logger.debug(f'decode frame #{index}') if self.metadata.BitsAllocated == 1: unpacked_frame = unpack_bits(frame_data) rows, columns = self.metadata.Rows, self.metadata.Columns n_pixels = self._pixels_per_frame pixel_offset = int(((index * n_pixels / 8) % 1) * 8) pixel_array = unpacked_frame[pixel_offset:pixel_offset + n_pixels] return pixel_array.reshape(rows, columns) frame_array = decode_frame( frame_data, rows=self.metadata.Rows, columns=self.metadata.Columns, samples_per_pixel=self.metadata.SamplesPerPixel, transfer_syntax_uid=self.metadata.file_meta.TransferSyntaxUID, bits_allocated=self.metadata.BitsAllocated, bits_stored=self.metadata.BitsStored, photometric_interpretation=self.metadata.PhotometricInterpretation, pixel_representation=self.metadata.PixelRepresentation, planar_configuration=getattr(self.metadata, 'PlanarConfiguration', None)) # We don't use the color_correct_frame() function here, since we cache # the ICC transform on the reader instance for improved performance. if correct_color and self._color_manager is not None: logger.debug(f'correct color of frame #{index}') return self._color_manager.transform_frame(frame_array) return frame_array @property def number_of_frames(self) -> int: """int: Number of frames""" try: return int(self.metadata.NumberOfFrames) except AttributeError: return 1
def _build_bot(fp: DicomFile, number_of_frames: int) -> List[int]: """Builds a Basic Offset Table (BOT) item of an encapsulated Pixel Data element. Parameters ---------- fp: pydicom.filebase.DicomFile Pointer for DICOM PS3.10 file stream positioned at the first byte of the Pixel Data element following the empty Basic Offset Table (BOT) number_of_frames: int Total number of frames in the dataset Returns ------- List[int] Offset of each Frame item in bytes from the first byte of the Pixel Data element following the BOT item Note ---- Moves the pointer back to the first byte of the Pixel Data element following the BOT item (the first byte of the first Frame item). Raises ------ IOError When file pointer is not positioned at first byte of first Frame item after Basic Offset Table item or when parsing of Frame item headers fails ValueError When the number of offsets doesn't match the specified number of frames """ initial_position = fp.tell() offset_values = [] current_offset = 0 i = 0 while True: frame_position = fp.tell() tag = TupleTag(fp.read_tag()) if int(tag) == SequenceDelimiterTag: break if int(tag) != ItemTag: fp.seek(initial_position, 0) raise IOError( 'Building Basic Offset Table (BOT) failed. ' f'Expected tag of Frame item #{i} at position {frame_position}.' ) length = fp.read_UL() if length % 2: fp.seek(initial_position, 0) raise IOError('Building Basic Offset Table (BOT) failed. ' f'Length of Frame item #{i} is not a multiple of 2.') elif length == 0: fp.seek(initial_position, 0) raise IOError('Building Basic Offset Table (BOT) failed. ' f'Length of Frame item #{i} is zero.') first_two_bytes = fp.read(2, 1) if not fp.is_little_endian: first_two_bytes = first_two_bytes[::-1] # In case of fragmentation, we only want to get the offsets to the # first fragment of a given frame. We can identify those based on the # JPEG and JPEG 2000 markers that should be found at the beginning and # end of the compressed byte stream. if first_two_bytes in _START_MARKERS: current_offset = frame_position - initial_position offset_values.append(current_offset) i += 1 fp.seek(length - 2, 1) # minus the first two bytes if len(offset_values) != number_of_frames: raise ValueError( 'Number of frame items does not match specified Number of Frames.') else: basic_offset_table = offset_values fp.seek(initial_position, 0) return basic_offset_table