The project consists of methods to build a data set from patient-specific samples, providing the user with a tidy, organized table to explore and hypothesize. In explicit, from a given set of .xml files exported from the EchoPAC software, a single table is generated, consisting of all relevant fields accounted for during clinical image analysis. With cases as rows and features as columns, the data set is designed for quick exploration and statistical research.
From the EchoPAC exports, the following parameters are extracted:
- Heart rate,
- Valve opening times,
- Blood pressure,
- Ejection fraction
- Global and segmental work indices:
- Myocardial work efficiency,
- Constructive work,
- Wasted work,
- Positive work,
- Negative work,
- Systolic constructive work,
- Sysystolic wasted work,
- Post-systolic strain,
- Average segmental strain,
- Systolic and diastolic pressure,
- Global and segmental strain measurements.
In addition to the raw indices, new features were derived, to provide the user with information relevant for results publication, such as:
- Frame rate for each of the views (4C, 3C, 2C),
- Segmental post-systolic boolean classification, which indicates whether the minimum strain took place after the aortic valve closure (AVC),
- Segmental post-systolic index - ratio between the strain at AVC and the minimum strain,
- Segmental minimum strain value,
- Segmental strain value at AVC,
- Segmental time-to-peak - time from th beginning of the cycle to minimum strain,
- Segmental time-to-peak ratio - ratio between time-to-peak and cycle duration,
- Minimum global strain before AVC,
- Minimum global strain,
- Time of minimum global strain.
A few additional functions are implemented to ease the visualization and comparative studies. In explicit, the values necessary for createing 17 and 18 AHA polar plots of the left ventricle are provided. The functions build a data frame consisting of mean and median values of the parameters of interest, one for each of the 17 or 18 segments. These values can be obtained for multiple categorized patient groups. Moreover, the representatives of each group (closest to the mean or median with regards to segmental values) are found.
Statistical models rely on the sets of samples to infer the behaviour of similar samples, classifiy them and build predictions based on them. In general, the more samples are gathered, the more robust and reliable the model is and more features of a sample can be explored. Population studies on the relatively new indices of function such as strain and myocardial work are becoming increasingly popular, however it is difficult to gather the data necessary for analysis and inductive reasoning.
EchoPAC software enables an export of individual patient data on the structure and function of the left ventricle. The data is saved in the .xml or .txt files, which are not designed for research. Obtaining comparable data is hindered by the structure of these exports, making it difficult to perform comparative studies for clinicians. With this tool, the data is transformed into easily applicable structure at a minimal cost.
All presented screenshot come form randomly created, probable data representing the original exports from the EchoPAC software.
The data is provided in a specific XML structure, which is illegible. It is also possible to open the file with applications such us MS Excel, as a table. However, population studies on these files would be time-consuming and laborious.
Less data is available in the case of a single view export. However, it is still possible to extract the data on segmental strain (color-coded).
| ID | HR (1/min) | MVC (ms) | ... | GWE (%) | GWI (mmHg%) | ... | SBP (mmHg) | DBP (mmHg) | MW_Basal Inferior (mmHg%) | MW_Basal Posterior (mmHg%) | ... |
|---|---|---|---|---|---|---|---|---|---|---|---|
| ABC001 | 63 | 24 | ... | 0.97 | 2054 | ... | 136 | 83 | 1524 | 2234 | ... |
| ACD002 | 77 | 45 | ... | 0.99 | 1725 | ... | 121 | 73 | 1281 | 1875 | ... |
| ADE003 | 79 | 38 | ... | 0.97 | 2262 | ... | 152 | 92 | 1362 | 2893 | ... |
| AEF004 | 65 | 32 | ... | 0.95 | 1254 | ... | 115 | 77 | 1123 | 1490 | ... |
| AFG005 | 67 | 13 | ... | 0.96 | 2032 | ... | 134 | 85 | 1552 | 1750 | ... |
| AGH006 | 84 | 37 | ... | 0.98 | 2472 | ... | 126 | 72 | 2762 | 3312 | ... |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| statistic_parameter_ group label | Basal Inferior | Basal Posterior | Basal Lateral | Basal Anterior | Basal Anteroseptal | Basal Septal | Mid Inferior | ... |
|---|---|---|---|---|---|---|---|---|
| mean_MW_2 | 1878 | 2243 | 2211 | 1855 | 1960 | 1739 | 2088 | ... |
| median_MW_2 | 1931 | 2234 | 2179 | 1855 | 1931 | 1693 | 2088 | ... |
| mean_MW_1 | 1753 | 2065 | 2157 | 1834 | 1710 | 1770 | 1997 | ... |
| median_MW_1 | 1734 | 2056 | 2275 | 1843 | 1705 | 1770 | 1997 | ... |
| mean_strain_avc_2 | -17 | -18 | -19 | -17 | -17 | -15 | -20 | ... |
| median_strain_avc_2 | -17 | -17 | -19 | -17 | -18 | -14 | -20 | ... |
| mean_strain_avc_1 | -18 | -16 | -19 | -18 | -18 | -17 | -22 | ... |
| median_strain_avc_1 | -19 | -16 | -18 | -17 | -18 | -17 | -21 | ... |
| Label | MW | strain_avc | strain_min | all |
|---|---|---|---|---|
| 0 | AAA0 | AAA1 | AAA2 | AAA1 |
| 1 | BBB0 | BBB1 | BBB2 | BBB3 |
| 2 | CCC0 | CCC1 | CCC2 | CCC0 |
Call
from echo_data_set import EchoDataSet
path_to_data = 'data/exports'
path_to_output = 'data/output'
eds = EchoDataSet(input_path=path_to_data, output_path=path_to_output, output='all_cases.csv',
export_file_type='xml', timings_file='timings.xlsx')
Input
input_path: path to the folder with .xml/.txt files containing the exports from EchoPAC;
output_path: path to a folder where the resulting table, the 17 and/or 18 AHA values and group representatives will be stored;
output: name of the file to which the table is saved;
export_file_type: xml or txt, the type of exports from which the data set is created;
timings_file: used only in the case of txt exports. The file should contain the timings of AVC in milliseconds for each available patient.
Output
An .xlsx or .csv file with the data set of all patients available in the input folder.
build_data_set_from_xml_files()Creates the popluation data from the EchoPAC exports in xml format (with myocardial work indices).
build_data_set_from_txt_files()Creates the popluation data from the EchoPAC exports in txt format (from single view).
get_aha_values(label_col='Classification', n_segments=18, labels_file='List of patients with labels.xlsx',
representatives=True)Obtain the mean and median segmental values of parameters of interest, with respect to the patient classes (groups).
Parameters:
- features: parameters to extract
- label_col: name of the column with classification of the patients
- representatives: whether to produce the table of class (group) representative patients
- n_segments: 17 or 18 - as used for creating the AHA plots
- labels_file: name of the file containing patient classification
Output: Either the list of group representatives or means and medians segmental values of the parameters of interest
Please quote the following publication:
Loncaric F, Nunno L, Mimbrero M, Marciniak M, Fernandes JF, Tirapu L, Fabijanovic D, Sanchis L, Doltra A, Cikes M, Lamata P, Bijnens B, Sitges M. Basal Ventricular Septal Hypertrophy in Systemic Hypertension. Am J Cardiol. 2020 May 1;125(9):1339-1346. doi: 10.1016/j.amjcard.2020.01.045.
The code is openly available. If this tool has been useful in your research, please reference this site https://github.com/MaciejPMarciniak/EchopacExportsReader