DCMTK解析DICOM汇总

最新推荐文章于 2025-04-02 10:37:44 发布

置顶 reikocao

最新推荐文章于 2025-04-02 10:37:44 发布

阅读量2.1k

点赞数

本文链接：https://blog.youkuaiyun.com/u014015324/article/details/110679783

版权

本文介绍DICOM文件格式的解析方法及CT图像处理流程，包括如何从DICOM文件中提取像素数据、计算CT值，并应用窗宽窗位进行图像过滤与显示。此外还涉及了DICOM图像压缩解压的具体实现。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

DICOM头信息示例如下：

(0002,0000) Group Length                                        VR: UL  Length: 4    Value: 184 
(0002,0001) File Meta Information Version                       VR: OB  Length: 2    Value: 0
(0002,0002) Media Storage SOP Class UID                         VR: UI  Length: 26   Value: 
(0002,0003) Media Storage SOP Instance UID                      VR: UI  Length: 54   Value: 
(0002,0010) Transfer Syntax UID                                 VR: UI  Length: 18   Value: 
(0002,0012) Implementation Class UID                            VR: UI  Length: 24   Value: 
(0002,0013) Implementation Version Name                         VR: SH  Length: 8    Value: DCIE 2.2
(0008,0005) Specific Character Set                              VR: CS  Length: 10   Value: ISO_IR 192
(0008,0008) Image Type                                          VR: CS  Length: 22   Value: ORIGINAL\PRIMARY\AXIAL
(0008,0012) Instance Creation Date                              VR: DA  Length: 8    Value: 20181115
(0008,0013) Instance Creation Time                              VR: TM  Length: 6    Value: 151330
(0008,0016) SOP Class UID                                       VR: UI  Length: 26   Value: 
(0008,0018) SOP Instance UID                                    VR: UI  Length: 54   Value: 1.2.840.113619.2.278.3.346865037.388.1539042181.XXX.XX
(0008,0020) Study Date                                          VR: DA  Length: 8    Value: 20181009
(0008,0021) Series Date                                         VR: DA  Length: 8    Value: 20181009
(0008,0022) Acquisition Date                                    VR: DA  Length: 8    Value: 20181009
(0008,0023) Image Date                                          VR: DA  Length: 8    Value: 20181009
(0008,0030) Study Time                                          VR: TM  Length: 6    Value: 111141
(0008,0031) Series Time                                         VR: TM  Length: 6    Value: 111308
(0008,0032) Acquisition Time                                    VR: TM  Length: 10   Value: 111515.407
(0008,0033) Image Time                                          VR: TM  Length: 6    Value: 111520
(0008,0050) Accession Number                                    VR: SH  Length: 0    
(0008,0060) Modality                                            VR: CS  Length: 2    Value: CT
(0008,0070) Manufacturer                                        VR: LO  Length: 18   Value: GE MEDICAL SYSTEMS
(0008,0080) Institution Name                                    VR: LO  Length: 26   Value: 
(0008,0090) Referring Physician's Name                          VR: PN  Length: 8    Value:  
(0008,1010) Station Name                                        VR: SH  Length: 4    Value: rtct
(0008,1030) Study Description                                   VR: LO  Length: 2    Value: HN
(0008,103E) Series Description                                  VR: LO  Length: 12   Value: CHEST 2.5 MM
(0008,1048) Physician(s) of Record                              VR: PN  Length: 14   Value:  
(0008,1070) Operators' Name                                     VR: PN  Length: 14   Value: 
(0008,1090) Manufacturer's Model Name                           VR: LO  Length: 18   Value: Discovery CT590 RT
(0008,1140) Referenced Image Sequence                           VR: SQ  Length: 104  
    (FFFE,E000) 
        (0008,1150) Referenced SOP Class UID                            VR: UI  Length: 26   Value: 
        (0008,1155) Referenced SOP Instance UID                         VR: UI  Length: 54   Value: 
(0008,3010) Irradiation Event UID                               VR: UI  Length: 52   Value: 
(0010,0010) Patient's Name                                      VR: PN  Length: 14   Value: 
(0010,0020) Patient ID                                          VR: LO  Length: 8    Value: XXXXX
(0010,0030) Patient's Birth Date                                VR: DA  Length: 8    Value: 
(0010,0032) Patient's Birth Time                                VR: TM  Length: 6    Value: 
(0010,0040) Patient's Sex                                       VR: CS  Length: 2    Value:  
(0018,0022) Scan Options                                        VR: CS  Length: 12   Value: HELICAL MODE
(0018,0050) Slice Thickness                                     VR: DS  Length: 4    Value: 2.5 
(0018,0060) KVP                                                 VR: DS  Length: 4    Value: 120 
(0018,0090) Data Collection Diameter                            VR: DS  Length: 4    Value: 500 
(0018,1000) Device Serial Number                                VR: LO  Length: 2    Value: * 
(0018,1020) Software Version(s)                                 VR: LO  Length: 10   Value: rt_bjcl.35
(0018,1100) Reconstruction Diameter                             VR: DS  Length: 4    Value: 650 
(0018,1110) Distance Source to Detector                         VR: DS  Length: 8    Value: 1062.55 
(0018,1111) Distance Source to Patient                          VR: DS  Length: 8    Value: 605.945 
(0018,1120) Gantry/Detector Tilt                                VR: DS  Length: 2    Value: 0 
(0018,1130) Table Height                                        VR: DS  Length: 4    Value: 174 
(0018,1140) Rotation Direction                                  VR: CS  Length: 2    Value: CW
(0018,1150) Exposure Time                                       VR: IS  Length: 4    Value: 856 
(0018,1151) X-Ray Tube Current                                  VR: IS  Length: 2    Value: 99
(0018,1152) Exposure                                            VR: IS  Length: 2    Value: 13
(0018,1160) Filter Type                                         VR: SH  Length: 12   Value: BODY FILTER 
(0018,1170) Generator Power                                     VR: IS  Length: 6    Value: 84000 
(0018,1190) Focal Spot(s)                                       VR: DS  Length: 4    Value: 1.2 
(0018,1210) Convolution Kernel                                  VR: SH  Length: 8    Value: STANDARD
(0018,5100) Patient Position                                    VR: CS  Length: 4    Value: HFS 
(0020,000D) Study Instance UID                                  VR: UI  Length: 52   Value: 1.2.840.113619.2.278.3.346865037.388.15390421XX.XXX
(0020,000E) Series Instance UID                                 VR: UI  Length: 52   Value: 1.2.840.113619.2.278.3.346865037.388.15390421XX.XXX
(0020,0010) Study ID                                            VR: SH  Length: 6    Value:  XXXXX
(0020,0011) Series Number                                       VR: IS  Length: 2    Value: 2 
(0020,0012) Acquisition Number                                  VR: IS  Length: 2    Value: 1 
(0020,0013) Image Number                                        VR: IS  Length: 2    Value: 19
(0020,0032) Image Position (Patient)                            VR: DS  Length: 16   Value: -325\-325\157.5 
(0020,0037) Image Orientation (Patient)                         VR: DS  Length: 12   Value: 1\0\0\0\1\0 
(0020,0052) Frame of Reference UID                              VR: UI  Length: 60   Value: 
(0020,1040) Position Reference Indicator                        VR: LO  Length: 0    
(0020,1041) Slice Location                                      VR: DS  Length: 6    Value: 157.5 
(0028,0002) Samples per Pixel                                   VR: US  Length: 2    Value: 1 
(0028,0004) Photometric Interpretation                          VR: CS  Length: 12   Value: MONOCHROME2 
(0028,0010) Rows                                                VR: US  Length: 2    Value: 512 
(0028,0011) Columns                                             VR: US  Length: 2    Value: 512 
(0028,0030) Pixel Spacing                                       VR: DS  Length: 18   Value: 1.269531\1.269531 
(0028,0100) Bits Allocated                                      VR: US  Length: 2    Value: 16 
(0028,0101) Bits Stored                                         VR: US  Length: 2    Value: 16 
(0028,0102) High Bit                                            VR: US  Length: 2    Value: 15 
(0028,0103) Pixel Representation                                VR: US  Length: 2    Value: 1 
(0028,1050) Window Center                                       VR: DS  Length: 2    Value: 40
(0028,1051) Window Width                                        VR: DS  Length: 4    Value: 400 
(0028,1052) Rescale Intercept                                   VR: DS  Length: 6    Value: -1000 
(0028,1053) Rescale Slope                                       VR: DS  Length: 2    Value: 1 
(0028,1054) Rescale Type                                        VR: LO  Length: 2    Value: HU

1，CT值

CT图像的CT 值反映组织对X 射线吸收值（衰减系数u），其单位为Hounsfield Unit（Hu），以水的衰减系数为参照（即水的CT 值为0）；物质衰减系数大于水者为正值，小于水者为负值；并以骨皮质和空气的衰减系数为上限和下限，定为＋1000 和－1000。

一般CT图像是经过量化后的灰度图像，标准的CT 灰度图像为12 位灰度图象。

获取CT值的公式位 Rescale slope∗pixelValue+Rescale intercept，其中pixelValue是从dicom文件中读取的像素值，得到CT值然后根据窗宽窗位过滤，过滤后得到的图像将凸显出所关注的目标组织。

2，截距，Rescale intercept (0028|1052) ，斜率，Rescale slope (0028|1053)

CT图像的像素值是用Hounsfield单位来测量的，而Hounsfield单位可以是负值，而CT图像的像素值通常存储为无符号整数。因此CT DICOM文件通常具有负截距。

对于线性缩放同样适用，一个像素可能有一个大范围的值，而存储的值需要占用尽可能少的bit，同时避免量化错误，这通常应用于PET成像，在PET成像中，所测量的活度值的范围可能超过小整数。

此外PET的最大活性可能因切片而异，例如与健康组织切片相比，带有肿瘤的切片可能具有很高的值。由于这个原因，Rescale intercept/Rescale slope在一个PET图像系列的切片基础上经常变化。

所以针对CT和PET，需要将实际值利用斜率和截距进行线性变换然后存储到DICOM文件中，其他情况斜率默认为1，截距默认为0。

3，DCM_PixelRepresentation

像素数据的存储类型，0代表无符号存储，1代表有符号存储

4，DCM_BitsAllocated

存储一个像素分配的bit数

4，DCM_BitsStored

DCM_BitsAllocated中的有效存储位，即有效数据所占字节长度

5，DCM_SamplesPerPixel

像素采样率，1表示灰度图，3表示彩色图

6，窗宽窗位

CT等放射医学影像常有窗宽（Window Width，简写WW）窗位（Window Level，简写WL）的转换。常规显示器颜色位深是8bit，彩色显示器是RGB每通道位深8bit，而医学图像通常是10~12bit（通常用16bit的变量类型表示）。

因此即使不做窗宽窗位转换也需要将10~12bit的数据映射到8bit来显示（显示器是彩色，则灰度图像的话每通道颜色一样即可）。

常见人体组织的CT值（HU）：

组织          CT值            组织        CT值
骨组织      ＞400           肝脏      50～70
钙值       80～300         脾脏      35～60
血块       64～84           胰腺      30～55
脑白质     25～34         肾脏      25～50
脑灰质     28～44         肌肉      40～55
脑脊液      3～8            胆囊      10～30
血液       13～32          甲状腺   50～90
血浆       3～14            脂肪  －20～－100
渗出液      ＞15           水           0

1、胸部CT检查时，肺窗纵膈窗窗宽、窗位分别是：

（1）肺窗WW1500—2000HU 、WL-450—-600HU

（2）纵膈窗WW250—350HU、WL30—50HU

2、骨窗、软组织窗窗宽、窗位

（1）骨窗WW1000—1500HU、WL250—350HU

（2）软组织窗WW300—500HU、WL40—60HU

假设要将CT图中的某组织在显示器上凸显出来，且该组织的WW为400，WL为60，则该组织对应的CT值范围为-140~260，其中WW为400，表示以0为界限，-200~200，WL为60，表示偏移60，得到-140~260。

从dicom文件中解析出像素数据，计算出原始CT值，然后根据WW和WL过滤，将过滤后的像素值归一化，然后通过opencv显示图片

7，使用DCMTK解压DICOM

根据DCM_TransferSyntaxUID得到dicom的传输语法，然后进行解压，代码如下

void DcmParse::Decompress(DcmDataset* dataset, const char* transferSyntax)
{
	std::string losslessTransUID = "1.2.840.10008.1.2.4.70";
	std::string lossTransUID = "1.2.840.10008.1.2.4.51";
	std::string losslessP14 = "1.2.840.10008.1.2.4.57";
	std::string lossyP1 = "1.2.840.10008.1.2.4.50";
	std::string lossyRLE = "1.2.840.10008.1.2.5";

	E_TransferSyntax xfer = dataset->getOriginalXfer();
	if (transferSyntax == losslessTransUID || transferSyntax == lossTransUID ||
		transferSyntax == losslessP14 || transferSyntax == lossyP1)
	{
		DJDecoderRegistration::registerCodecs();
		dataset->chooseRepresentation(EXS_LittleEndianExplicit, NULL);
		DJDecoderRegistration::cleanup();
	}
	else if (transferSyntax == lossyRLE)
	{
		DcmRLEDecoderRegistration::registerCodecs();
		dataset->chooseRepresentation(EXS_LittleEndianExplicit, NULL);
		DcmRLEDecoderRegistration::cleanup();
	}
	else
	{
		dataset->chooseRepresentation(xfer, NULL);
	}
}

如果要使用DicomImage类，则可以（xfer为传输语法）：

DicomImage* dcmImage = new DicomImage((DcmObject*)dataset, xfer);

根据dicom路径获得像素数据代码如下：

    DcmFileFormat fileformat;
	OFCondition oc = fileformat.loadFile(dcmPath.c_str());
	if (!oc.good()) {
		return FILE_NOT_GOOD;
	}
	const char* transferSyntax = NULL;
	DcmDataset* dataset = fileformat.getDataset();
	//获取传输语义
	fileformat.getMetaInfo()->findAndGetString(DCM_TransferSyntaxUID, transferSyntax);
	std::cout << "transferSyntax is " << transferSyntax << std::endl;
	//解压
	Decompress(dataset, transferSyntax);
	//判断文件是否损坏
	DcmElement* element = NULL;
	OFCondition result = dataset->findAndGetElement(DCM_PixelData, element);
	if (result.bad() || element == NULL) {
		return PIXEL_DATA_IS_BAD;
	}
    //dataset根据dicom的通道数以及数据存储位数获取，略