Walk through directory then find a text file in specifically named subdirectories and extract data to a list

Question

I have a parent directory named 'scratch', within that parent dir I have several folders with random names, then within those folders there is a folder named 'deagg', then within 'deagg' there is a random folder name, then within that there is a 'summary.txt' file that I want to parse through to get numbers out of. If a line in the 'summary.txt' file says 'Slab' this extracts the float number in that line, then adds it to the list slab_sum.

My issue is with the small code below. The only way I could get it to append the total of 'slab' to the list was to detect a change in directory. How can I have this append the slab_total to slab_sum list for each directory/summary.txt without using the small code below to detect directory change. I want it to parse through the summary.txt file, then once it does that send it to the slab_sum list. Also, I have four of these...not just slab, I made it simple for this example, so there are four different key words I search for in each summary.txt.

if prev_dir is None:
    prev_dir = root
elif prev_dir != root:
    slab_sum.append(str(slab_total))
    prev_dir = root

Example directory: 'C://scratch//PGA//deagg//apple/summary.txt'

File = 'summary.txt'
Start_Path = 'C://scratch'
slab_sum = []

def extract_data():
    prev_dir = None
    for root, dirs, files in os.walk(Start_Path):
        if File in files and root.split(os.path.sep)[1] == "deagg":
            summary_path = root + os.path.sep + File
            search_file = open(summary_path, 'r', encoding='utf-8')
            slab_count = 0
            slab_number = []
            for line in search_file:
            #for i, line in enumerate(search_file):
                # Parse through 'summary.txt' for any line that has 'Slab', record only the end float number.
                # Convert float to decimal rounded to hundredth place
                if "Slab" in line:
                    slab_num = ([float(s) for s in re.findall(r'[-+]?(?:\d*\.\d+|\d+)', line)])
                    slab_number.append(slab_num)
                    slab_count = slab_count + 1
                    slab_total = 0
                    for slab_num in slab_number:
                        slab_total += slab_num
                if prev_dir is None:
                    prev_dir = root
                elif prev_dir != root:
                    slab_sum.append(str(slab_total))
                    prev_dir = root
            search_file.close()